This presentation is confidential.
Please enter your access code to continue.
1 / auto
⚠ Confidential — Authorised Parties Only
Weavi-QCE

The infrastructure layer
that makes enterprise AI economically viable

auto
The investment thesis — one slide

A selection layer
for AI workloads that outgrow direct context.

The problem

Enterprise corpora can exceed model context windows by orders of magnitude. Sending, chunking or repeatedly summarising everything adds cost, latency and information risk before reasoning begins.

The solution

The QCE — Quantum Context Engine applies a proprietary quantum-inspired selection framework to configured record signals and a fixed token budget. It returns full selected records—not generated summaries—for the customer's downstream model.

The proof

Six public semantic domains, 898 independently labelled query units, 21,130 recorded arm-level outcomes, 2,574 model/planner requests and physical testing through 5M estimated tokens. Failures remain reviewable.

Evidence date: 15 September 2026 · Models tested: Claude Sonnet 5 · Gemini 3.8 Flash · Claude Opus 5 · Datasets: 6 public domains, real SEC EDGAR, OpenFDA
96.5%
Recall · 500K→256K
91.5%
Recall · physical 5M→1M
80.0%
Input removed at physical 5M
~50%
Median paired 500K model-token saving
Where QCE sits
📁
Enterprise data
Any volume
Files · emails · records · systems
all records
Weavi QCE
Evidence selection
Deterministic · Provenance-aware
Removes 50–99% of input
evidence budget only
🧠
Your AI model
Focused reasoning
Any LLM · Sonnet, Gemini, Opus…
~50–99%
Observed range · varies by workload
892 data points · Sep 2026
circa 50%
Observed model-token saving · 6–8 pairs
Sonnet 5 · Gemini 3.8 · Opus 5
up to 87–97%
Observed cost reduction · 1 valid pair each
Real SEC EDGAR diagnostics
up to 91.5%
Evidence recall in our tests · physical 5M
With around 80% input reduction

The latest result is a measurable infrastructure wedge: QCE processed a 5M-token physical corpus that the tested full-model arms could not accept, retained 91.5% of labelled evidence, and passed only 1M tokens onward. Partner-owned replication is the next value inflection.

See evidence scope and caveats
898independently labelled query units
6 domainsscience, health, finance, policy, public interest and climate
0/10full 5M model calls eligible in the latest three-model study
62.5%QCE recall at represented 500M→5M—the open R&D frontier

Arm-level outcome counts include ablations, scale points and model comparisons; they are not independent samples. Physical 5M is distinct from represented 500M. The latest research candidate is not automatically routed in production.

3 / auto
How it works — overview

Three steps.
One measurable selection layer.

01 — INGEST

Many sources, one evidence budget

Supply authorised records, messages or document chunks with stable IDs, source/speaker metadata and token costs. Very large archives use bounded shards rather than one unsafe request.

02 — FIND

Quantum-inspired selection

Our algorithm scores every record across multiple mathematical dimensions simultaneously using our proprietary quantum-inspired framework.

03 — RETURN

Exact data. Not summaries.

We return the actual records — unmodified, unparaphrased. Your LLM gets the truth, not an interpretation of it.

The latest scale ladder reported 48.8% reduction with 96.5% recall at 500K, and 80.0% reduction with 91.5% recall at physical 5M. At represented 500M→5M, reduction reached 99.0% but recall fell to 62.5%—the next quality frontier.

See the three-scale metric table
Input → budgetRecallPrecisionMAP / MRRReduction
500K → 256K96.5%1.39%.5387 / .688048.8%
Physical 5M → 1M91.5%.80%.3996 / .559980.0%
Represented 500M → 5M62.5%12.48%.3918 / .557699.0%

Precision uses incomplete public qrels: unjudged selected records count as nonrelevant. Represented 500M reweights a bounded physical 5M corpus; it is not physical ingestion of 500M tokens.

auto
The architecture — visual

Two paths.
One measurable control point.

Without Weavi-QCE — the corpus exceeds the tested model boundary
Your data
5,000,000
tokens
Latest tested LLMs
Ineligible
1.0–1.048M contexts
Result
No direct run
0/3 model families eligible

In the locked no-assistance study, full physical-5M inputs were context-ineligible for Sonnet 5, Gemini 3.8 Flash and Opus 5. Ineligibility is an operational outcome—not a model-quality score.

✓ Mode A — Weavi-QCE + Customer's LLM
Your data
5,000,000
tokens
Weavi-QCE API
QCE
api.weavi.ai
Selected evidence
~1,000,000
tokens · 80% reduction
Result
91.5% recall
algorithm-only labelled evidence

Customer keeps model choice. QCE returned a deterministic 1M-token evidence budget from 5M physical tokens. Gemini 3.8 accepted eight of ten reduced semantic prompts; Sonnet 5 and Opus 5 remained ineligible after declared input plus overhead.

See model-boundary evidence

500K: all three latest model families were eligible for an attempt; each completed 8/10 semantic calls. 5M: no full-model arm was eligible. Represented 500M: QCE reduced declared input by 99%, but the approximately 5M-token output remained beyond every tested context.

At matched 500K valid pairs, QCE reduced model tokens by a median 53.17%. Recall improved for Sonnet and Opus, and fell 1.25 points for Gemini. Pair counts were 6–8, so this is diligence evidence rather than formal superiority.

✦ Mode B — Weavi-QCE + Weavi LLM (future revenue expansion)
Your data
1,000,000
tokens
Weavi-QCE API
QCE
api.weavi.ai
Weavi LLM
Domain-trained
vertical-specific
Result
Full stack
owned end-to-end

Future path: Weavi trains and operates domain-specific LLMs (legal, medical, financial) optimised to work natively with the QCE. Full-stack AI service. Higher margin. Deeper customer lock-in.

auto
"15 September 2026 · Physical 5M scale · Claude Sonnet 5, Gemini 3.8 Flash, Claude Opus 5 tested: Weavi-QCE retained 91.5% of labelled evidence while removing 80% of input. The full tested model arms could not accept the corpus. That is a measurable infrastructure result—and a clear invitation for partner diligence."
auto
Market scale — verified 2026 data

The numbers are real.
And they are staggering.

These are not forecasts. These are verified figures already published in 2026. The AI token economy is here — and it is enormous.

Google — May 2026
3.2 quadrillion
tokens per month across Google products

7× year-on-year growth. 375+ Google Cloud customers each processed over 1 trillion tokens in the preceding 12 months. 19 billion tokens per minute via Google model APIs. Source: Google I/O keynote, 19 May 2026.

Gartner — May 2026 forecast
$2.60tn
worldwide AI spending in 2026

Rising to $3.49tn in 2027. AI model spending alone: $32.6bn in 2026 → $59.2bn in 2027 (+81%). By end of 2026, up to 40% of enterprise applications will include AI agents. Source: Gartner, May 2026.

U.S. Government — 2026
$90.7B
DoD potential AI contract value

28 U.S. federal agencies hold AI contracts. DoD alone: $90.7B potential value, 1,319 contracts. Alphabet Q1 2026: $35.7B capex, majority for AI infrastructure. Source: Brookings, May 2026; Alphabet IR.

Why this matters for Weavi-QCE
The problem is getting worse
Token volume is growing rapidly and every model-input token carries cost, latency and governance surface. The avoidable share is workload-specific and must be measured—not assumed.
The market is accelerating
AI model spending doubles every 18 months. By 2028, Gartner predicts 50%+ of customer-service organisations will double technology spending without equivalent staff reduction. Weavi-QCE prevents that.
The window is now
Enterprises are locking in AI infrastructure decisions in 2026. The organisation that solves token waste becomes the default layer for the entire AI economy — like AWS became the default for cloud.
auto
The $50B+ problem — verified 2026 AI budget data

The largest organisations
are already spending at model-input scale.

When QCE reduces model input by 50–99%, the financial saving scales with each organisation's actual AI spend. These are verified published figures, not projections.

JP Morgan Chase — Banking
$4.5B+
AI-specific 2026 technology budget
At our observed 50–97% test range, up to $2.25B–$4.37B illustrative potential saving

Across loan processing, fraud detection, compliance review, trade surveillance and customer intelligence — all text-heavy, query-intensive workloads.

Source: JPMorgan Chase 2025 Annual Report; $17.6B total tech budget with $4-5B AI-explicitly stated.

UK NHS — Healthcare
£1.5B+
NHS England AI & digital transformation budget
Up to £750M–£1.4B illustrative potential saving

Patient records, clinical notes, imaging reports and diagnostic history — massive, distributed, multi-source corpora with strict zero-retention requirements.

Source: NHS England Long Term Workforce Plan 2025; NHSE Digital AI Strategy 2026.

US Federal Government — AI
$10B+
FY2026 federal AI expenditure (full government)
Up to $5B–$9.7B illustrative potential saving

28 agencies with AI contracts. DoD: $90.7B potential AI contract value. Immigration, benefits, tax compliance, intelligence — all text-heavy LLM inference workloads.

Source: OMB FY2026 Budget Request; Brookings Institution AI contracts analysis, May 2026.

Goldman Sachs — Investment Banking
$1.5B
Annual AI spend 2025–2026

Research, M&A analysis, risk modelling, trade surveillance. Up to $750M–$1.45B illustrative.

Source: GS Technology Strategy 2025 Investor Day.

EU Banking Sector — Combined
€5.4B
Cumulative AI investment by European banks

AML/KYC, credit risk, regulatory reporting. Up to €2.7B–€5.2B illustrative.

Source: ECB Financial Stability Review, AI in Banking, 2025.

Gartner — Global AI model spend
$32.6B
Global AI model API spend 2026 → $59.2B by 2027

At our observed 50–97% test range, up to $16B–$31.6B illustrative addressable saving globally in 2026.

Source: Gartner Worldwide AI Market Forecast, May 2026.

The opportunity framing: QCE is not competing for a share of the AI budget — it is positioned to reduce the largest single line item in that budget (inference token cost) for every organisation that runs LLMs on governed enterprise data. At scale, the total addressable saving exceeds the entire global AI market spend.

3 / 28
The founder

A founder-led framework
built at the intersection of physics, AI and security.

The QCE applies mathematical ideas developed through a long-running physics programme to budget-aware information selection, informed by decades of work in security and regulated environments.

The origin of the algorithm
  • → Physics background spanning decades — quantum mechanics, gravitational theory, information theory
  • → Developed an original quantum gravitational theory ~15 years ago
  • → Recognised that the same mathematical principles governing gravitational field selection apply to information relevance in large datasets
  • → Applied this framework to AI context selection — the result is the QCE algorithm family
  • → This is not a derivative of existing NLP approaches — it is a fundamentally different mathematical framework
AI & technology background
  • → Active in AI since 2020 — NVIDIA, ARM ecosystem engagement
  • → Built and validated the QCE algorithm family across 10+ variants
  • → Production-ready API infrastructure — live at api.weavi.ai
  • → Built configurable test routes spanning SEC, Hacker News, OpenFDA and custom needle datasets
Information & cyber security — 20+ years
  • → National Chairman for security industry body — government, law enforcement, industry
  • → Industry partners include Amazon, Cisco, and major technology organisations
  • → Meetings at Westminster — policy-level engagement with UK government
  • → Deep understanding of regulated industries: defence, government, healthcare, finance
  • → This is why the architecture prioritises ephemeral processing, controlled deployment and minimised content logging
Why this matters to investors

The potential moat is the combination of a differentiated mathematical framework, a growing algorithm portfolio, implementation knowledge and domain configuration. Patents, independent benchmarks and customer deployment will determine how defensible that moat becomes.

4 / 28
Why us. Why now.

The right person.
The right moment. The right technology.

Why this founder
  • → 20+ years in regulated, high-stakes environments
  • → Original quantum gravitational theory — not borrowed from academia
  • → Government and law enforcement relationships — the exact customers who need this most
  • → Built the entire system: algorithm, API, validation, documentation
Why this moment
  • → Google: 3.2 quadrillion tokens/month (May 2026) — and growing 7× year-on-year
  • → Enterprises moving from AI pilots to production — cost walls hit now
  • → CFOs demanding measured AI unit economics
  • → Evidence, customer learning and IP protection can build a defensible lead
Why this technology
  • → Not prompt engineering — a mathematical framework
  • → Not summarisation — exact, unmodified records
  • → Not a single algorithm — a family of 10+ variants
  • → Validated on real data, not synthetic benchmarks

The convergence of a physicist with 20 years of enterprise security experience, a novel mathematical framework, and the exact moment when enterprise AI costs become existential — this is not a coincidence. This is the right person, with the right technology, at the right time.

auto
The scale advantage

Large corpora.
Selection before inference.

Enterprise corpora frequently exceed a model's usable context. The current evidence ladder spans public six-domain labels, real SEC material, 500K and 5M physical-token studies, and bounded represented pressure through 500M logical records.

Tested at scale
5M physical
20M characters · 5,000 records

QCE retained 91.5% of labelled evidence with an 80% declared reduction. Physical padding volume and independent semantic diversity are reported separately.

What LLMs can actually read
0 / 3
latest full-model families eligible at 5M

Sonnet 5, Gemini 3.8 Flash and Opus 5 advertised 1.0–1.048M contexts. The locked full 5M arms were ineligible before a provider call.

QCE + any LLM
500M represented
99% reduction under extreme pressure

QCE retained 62.5% of labelled evidence at 500M→5M. This exposes both the scalable mechanics and the open extreme-compression quality frontier.

Why volume creates an opportunity
Enterprise data is massive

Hospitals, law firms, financial institutions and governments hold corpora far beyond a single model call. A reliable pre-inference selection layer could make more of that evidence economically usable.

A testable scaling hypothesis

Latest matched results are now available: 96.5% recall at 500K, 91.5% at physical 5M, and 62.5% at represented 500M. Performance is budget-dependent rather than assumed scale-invariant.

The bigger the data, the bigger the saving

If quality is preserved, avoided model input grows with excluded tokens. Partner query volume, model pricing and realised retention will determine the actual economic value.

See physical versus represented scale definitions

Physical 500K: 1,753 complete records; 500K estimated tokens; 256K budget. Physical 5M: 5,000 records; exactly 20M content characters; 1M budget. Represented 500M: the bounded physical 5M corpus with 100× declared record weights and a 5M budget.

Represented scale validates ranking and budget pressure; it does not claim physical ingestion of 500M distinct tokens. Genuine customer archives require the implemented asynchronous manifest/shard direction plus a measured sharded-versus-unsharded equivalence study.

6 / 28
The problem

Your AI is reading everything.
It should only read what matters.

Organisations — government, healthcare, finance, legal, defence — hold vast data repositories. A model should receive a traceable evidence budget rather than an unaffordable haystack. Weavi is building and measuring that control layer.

Without Weavi-QCE

Needle lost in the haystack

  • Candidate corpora range from compact case histories to archives far beyond one model call
  • LLM costs: $1.25–$40+ per interaction at scale (Claude Sonnet 5: $2.50/1M input tokens)
  • Context windows overflow — critical data gets truncated
  • Model hallucinates from noise and irrelevant records
  • $50B wasted annually on irrelevant context globally
  • The answer is buried — and often missed entirely
With Weavi-QCE

Only the needle. Every time.

  • API selects full records under an explicit downstream token budget
  • Observed reduction varies from close-budget retention to severe-compression selection
  • Selection retention and downstream model use are measured separately
  • Full, unmodified records returned — no summaries, no hallucination
  • Works with any LLM — no vendor lock-in
  • Zero data retention — ephemeral processing only

The LLM is not a search engine. It is an intelligence engine. Weavi-QCE gives it intelligence — not a haystack.

7 / 28
How it works

Find the needle.
Feed it to your AI.

01 — SUPPLY

Authorised candidate data

Send bounded records, documents or messages from one or many sources. A first-stage index remains necessary for genuinely unbounded archives.

02 — FIND

Quantum-inspired selection

Our algorithm scores every record across multiple mathematical dimensions simultaneously using our proprietary quantum-inspired framework.

03 — RETURN

Exact data. Not summaries.

We return the actual records — unmodified, unparaphrased. Your LLM gets the truth, not an interpretation of it.

04 — MEASURE

Quality and savings together

Latest observed reductions span 48.8% at 500K, 80.0% at physical 5M and 99.0% under represented-500M pressure—with recall reported beside each.

auto
Revenue expansion — the full-stack opportunity

From QCE to
full-stack AI platform.

Today: the Weavi QCE (Quantum Context Engine) is the intelligence layer — the most defensible position in the AI stack. Tomorrow: with a trained domain LLM, Weavi becomes the complete AI infrastructure for regulated industries.

Stage 1 — Now
QCE API

Customer brings their own LLM. Weavi-QCE filters the context. Revenue: per-token API pricing. Gross margin: 85%+. No LLM cost to us.

Revenue model: SaaS + enterprise custom
Stage 2 — 3–6 months
QCE + LLM

Weavi trains domain-specific LLMs (legal, medical, financial, defence) optimised to work natively with the QCE. Customer gets a complete, pre-integrated AI stack.

Revenue model: Platform fee + per-query pricing + revenue share
Stage 3 — 6–12 months
Full-stack AI

Weavi is the default AI infrastructure for regulated industries. Data in → intelligence out. No third-party LLM dependency. Highest margin. Deepest moat.

Revenue model: Annual enterprise contracts + outcome-based pricing
Why domain LLMs are the natural next step
  • → The QCE already understands what data matters in each vertical — it scores relevance using quantum mechanical mathematics
  • → Training a domain LLM on pre-filtered, high-signal data produces a better model at lower cost
  • → Regulated industries (healthcare, legal, defence) cannot use public LLMs — they need private, auditable models
  • → Weavi already has the compliance architecture (zero data retention, HIPAA, GDPR, on-premise)
The compounding moat

Each enterprise deployment generates performance data that improves QCE routing. A domain LLM trained on QCE-selected, high-signal data is inherently better than one trained on raw data. The more customers, the better the models. The better the models, the more customers. This is a compounding data flywheel.

8 / 28
Real-world integration demonstration

OpenFDA adverse events.
Two pipelines, one question.

The reference application queries public OpenFDA adverse-event reports, runs a selected LLM over a baseline cohort and over QCE-selected records, then displays both responses and token usage side by side. It demonstrates a working integration—not clinical validation.

How the demo works
Step 1 — Set parameters
You provide an OpenFDA search query and a maximum record count. The service fetches a real report cohort and assigns declared severity/detail importance signals.
Step 2 — Ask the question
You pose the same question to the same model under both conditions. Model and token budget are explicit request parameters.
Step 3 — Compare the results
Two full responses and their token usage are returned. Reviewers can inspect the efficiency difference; blinded factual-quality scoring is the next validation step.
Response A — LLM only (no QCE)

Query: "asthma teenagers — leading indicators"

  • Selected OpenFDA cohort formatted as full source records
  • Same user question and chosen model
  • Input/output token usage measured from the model response
  • May require truncation when the cohort exceeds model context
Response B — LLM + Weavi QCE

Same query, same LLM, same parameters

  • QCE selects a subset under the declared token budget
  • LLM receives full, unmodified selected report text
  • Same model, question and output settings
  • Token and cost differences are returned explicitly
Measured
Tokens, cost and latency

The endpoint returns the observed comparison for the chosen cohort, model and budget.

Side-by-side
Two model responses returned

Same question and model, different supplied context. No claim of clinical correctness is implied.

Reviewable
Parameters and source records declared

The next step is blinded scoring against a prespecified factual and citation rubric.

See observed model-token and cost economics
53.17%median paired model-token reduction across the latest 500K Sonnet/Gemini/Opus comparisons
87.22%paired model-token reduction in the 264K real SEC diagnostic
95.79%paired model-token reduction in the 880K real SEC diagnostic
74.81%model-token reduction in the independent 90-query QCE pipeline study

The real SEC medium/large economics each had only one valid full-model pair. In the 90-query independent study, savings traded away recall. Latest 500K paired valid-call recall improved for Sonnet and Opus and fell 1.25 points for Gemini; pair counts were only 6–8.

9 / 28
Statistical validation — the complete methodology

Use the tests the hypothesis needs.
Keep every outcome reviewable.

The Evidence Lab uses real SEC EDGAR 10-K records; the OpenFDA reference app uses live public adverse-event cohorts. Paired retrieval, ranking, resampling, uncertainty and multiplicity diagnostics are retained where mathematically applicable. Test count is not itself proof.

Hypothesis tests & significance
Paired t-test
Wilcoxon signed-rank test
Bonferroni correction
Benjamini-Hochberg FDR
McNemar's test
Permutation test
Test agreement analysis
Heterogeneity analysis
Effect size & confidence
Cohen's d effect size
Bootstrap CI (standard)
Bootstrap CI (BCa bias-corrected)
Bayesian credible interval
Normality & distribution
Anderson-Darling normality test
Comprehensive normality suite
Cross-validation & power
K-fold cross-validation
Leave-one-out CV
Pre-study power analysis
Achieved power calculation
Information theory
Information-theoretic bounds
Latency percentile analysis
Retrieval metrics — TREC standard (30+ years)
MAPMRRNDCG@3NDCG@5NDCG@10Precision@3Precision@5Precision@10Recall@3Recall@5Recall@10F1-ScorePrecisionRecall
Validation standards: FDA drug trials · Basel III banking · TREC benchmarks · Solvency II (99.5% CI)
Temperature: 0.0 · Seeds: published · Iterations: 30+ · Data: real SEC EDGAR 10-K filings
10 / 28
Verification & proof

We will prove it.
Live. With your data.

This is not a claim. It is a reproducible, auditable, verifiable fact. We operate on a zero-knowledge proof principle: you don't have to take our word for it. We will show you the test, run it live, and you can verify every result yourself.

What we will show you
  • ✓ The complete test source code
  • ✓ All statistical functions, unmodified
  • ✓ The exact seeds used for reproducibility
  • ✓ Live test execution — in real time
  • ✓ Real SEC EDGAR data — fetched live
What you can do
  • ✓ Bring your own data — we will run it
  • ✓ Choose the model — we will test it
  • ✓ Set the parameters — we will execute
  • ✓ Verify every statistical output yourself
  • ✓ Have your own statistician review the code
What we warrant
  • ✓ The test code is the test code — no hidden logic
  • ✓ Results are deterministic at temperature 0.0
  • ✓ Same seed = same result, every time
  • ✓ No cherry-picked runs — all 30+ iterations shown
  • ✓ We will sign an NDA before showing anything
The zero-knowledge proof principle

You give us your data. We run the algorithm. You see the token reduction, the quality scores, the statistical outputs — all calculated in front of you, with code you can inspect. You don't need to trust us. The mathematics speaks for itself.

11 / 28
Industry impact — illustrative use cases

Real industries.
Real potential. Illustrated scenarios.

Financial Services — SEC / M&A
$1M+
illustrative annual savings · 1,000 filings/day scenario

In large-corpus SEC filing scenarios: significant token reduction (typically 70–90%+ on dense financial documents). Cost/doc falls materially at Claude Sonnet 5 rates. Speed and accuracy both improve — exact results vary by dataset and query type.

Healthcare — Patient Records (EHR)
Significant
cost reduction potential · large EHR dataset scenario

Large patient-record datasets are a high-value validation target. Deployment would require customer-side recall testing, privacy review, contractual controls and applicable healthcare compliance—not only an internal benchmark.

Insurance — Claims & Fraud
Material
fraud detection improvement · multi-source claim scenario

Multi-source claim datasets (police reports, medical records, witness statements, policy docs) see significant token reduction. Cross-referencing across sources improves signal quality — fraud indicators that would be missed in single-source review are surfaced. Adjuster throughput increases materially.

Legal — E-Discovery
Up to 97%
cost reduction potential · large document corpus scenario

Large e-discovery corpora (1M+ emails/documents) benefit most — the QCE filters algorithmically before any LLM cost is incurred. Full corpus coverage replaces expensive sampling. Processing time reduces from weeks to hours. Results vary by corpus density and query specificity.

Banking — Basel III Compliance
$M–$M+
illustrative annual savings · large loan portfolio scenario

High-volume loan application processing benefits significantly from token reduction on dense regulatory documents. At scale (200K+ applications/year), savings compound materially at published LLM rates. Compliance review timelines reduce. Exact savings depend on LLM pricing, document density, and query volume.

Customer Support AI
Substantial
cost reduction · high-volume support scenario

Long conversation histories (thousands of messages) are where the QCE excels — selecting the relevant recent and historical context rather than sending everything. At high ticket volumes, per-ticket LLM cost falls dramatically. Exact savings depend on history length, LLM pricing, and query volume.

* Use-case financial figures are scenarios, not forecasts. Latest matched technical evidence observed 48.8% reduction with 96.5% recall at 500K and 80.0% reduction with 91.5% recall at physical 5M. Actual savings depend on model pricing, corpus density, query volume, selected budget and customer-side quality requirements.

12 / 28
High-impact use cases — the wow factor

From aircraft manuals
to mortgage finance and global government.

Aerospace — Major Aircraft Fleet
Transformative
potential impact · large maintenance manual corpus

25M+ token maintenance manuals are an ideal QCE use case — the corpus is vast, queries are specific, and the relevant answer is a tiny fraction of the total. Technician lookup time reduces dramatically. Engineer hours saved at scale. Safety risk reduced by surfacing the right procedure reliably.

Government — Tax Administration
High ROI
potential · large-scale tax screening scenario

150M+ tax returns represent a massive corpus where the QCE's scale advantage is most pronounced. Algorithmic pre-filtering at zero LLM cost, followed by semantic ranking, dramatically improves audit targeting precision. Hit rate improvement and cost reduction both compound at this scale.

Mortgage Finance — Underwriting & Compliance
80–95%
cost reduction potential · large mortgage portfolio scenario

A mid-tier bank processing 50,000 mortgage applications/year has 100M+ tokens of applications, valuations, credit reports, legal covenants and compliance documentation. QCE selects only the evidence relevant to the underwriting query — fraud signals, income verification, risk indicators — before any model call. Compliance review and credit-decision quality both improve.

Space — Launch Safety Analysis
Significantly faster
analysis time reduction · large launch data scenario

27M+ tokens of pre-launch data (telemetry, inspection reports, historical launches, weather) is a high-value QCE use case. The relevant safety signals are a small fraction of the total corpus. Analysis time and engineering resource requirements both reduce materially. Data coverage improves.

AI Platforms — Workspace Integration
$3.72B
ARR potential · major productivity platform

Average enterprise user: 300M tokens (emails + documents). Standard AI context window covers <1%. With Weavi-QCE: full archive fits. "Only AI that can search your entire work history." 4x adoption vs standard tier.

Legal AI — Case Law Research
Up to 98%+
cost reduction potential · large case law corpus scenario

Large case law corpora (500+ documents, 12M+ tokens) see the highest QCE savings — the relevant legal reasoning is a tiny fraction of the total. Single LLM pass replaces expensive multi-pass chunking. At scale across a legal practice, per-query cost falls dramatically. Results vary by corpus size and query specificity.

14 / 28
Who needs this — and why now

Four converging forces.
A timely infrastructure question.

Market timing — 4 forces converging now
1. Production adoption
Enterprises are moving from isolated demonstrations to recurring AI workflows, making input economics and reliability more visible.
2. Cost accountability
CFOs and product owners increasingly require measured AI unit economics rather than experimentation without a path to margin.
3. Model choice
A model-independent selection layer can help customers manage cost and context without making a single-model commitment.
4. Governance
Regulated workflows need traceable source records, controlled processing and measurable failure policies—not only fluent answers.
Buyer urgency — three tiers
Tier 1 — Platform leverage

AI and data platforms: one integration could expose context selection across many customer workloads, subject to independent benchmark and partner economics.

Tier 2 — Competitive advantage

Enterprise users: banks, healthcare networks, law firms, insurers and government teams with large, governed evidence corpora.

Tier 3 — Strategic enabler

AI product companies: teams with recurring long-context inference can test whether selection improves gross margin without reducing answer quality.

The timing is attractive, but the strategy should be disciplined: secure partner-owned evidence, prove one high-value workflow, protect the IP and let measured customer economics determine the pace and valuation.

15 / 28
Market entry

Start where context is
large, expensive and auditable.

We are not relying on a top-down percentage of global AI spend. The first market is a focused set of regulated, document-heavy workflows where token cost and retrieval quality can be measured against an existing baseline.

Commercial wedge

01
Prove one high-cost workflow
Customer corpus, agreed benchmark, measured savings and quality
02
Expand within the customer
More corpora, teams, queries and model providers
03
Distribute through platforms
Cloud, data and model partners can embed the selection layer
Sizing discipline

A defensible bottom-up market model will follow paid pilots: verified annual query volume × realised saving × value-based share. Until those inputs exist, large top-down revenue claims are scenarios, not forecasts.

Priority customer profiles

VerticalWhy the QCE fits
Financial servicesLong filings, auditability and high query volume
HealthcareLongitudinal records and strict data handling
Legal & insuranceLarge evidence corpora with measurable retrieval quality
Government & defenceMulti-source data, private deployment and corpus scale
Data & AI platformsDistribution leverage across many enterprise workloads
16 / 28
Competitive landscape

Why alternatives
fail at scale.

The relevant comparison is not a marketing percentage; it is recall, ranking quality, cost and latency under the same corpus, labels and token budget. Strong BM25, dense, hybrid and reranker baselines remain part of the confirmatory roadmap.

⚡ Weavi-QCE
49–99%*
Summarisation APIs
30–50%
TF-IDF
~60%
LLMLingua / Compression
~30%
BM25
~45%
Prompt Engineering
10–20%
Summarisation APIs
Costs tokens to compress

Introduces hallucination risk. Destroys original data fidelity. Not suitable for regulated industries. Loses the needle.

Build In-House
Time + evidence cost

A comparable layer requires algorithms, bounded execution, provider integrations, labelled evaluation and accumulated failure learning. Customer-specific build cost requires diligence.

Do Nothing
Unbounded input economics

Continue paying to send all context where models can accept it—and build ad hoc truncation where they cannot. The avoidable amount depends on each workload.

auto
Adversarial validation — tested against the best alternatives

We tested the hardest cases.
Against every known approach.

The adversarial needle test is designed to make retrieval as hard as possible — needles buried deep, surrounded by highly similar decoy content. We ran this against every major alternative approach. The results are not close.

Approach
Easy
Medium
Hard
Token saving
Weavi QCE
F1: 0.94
F1: 0.82
F1: 0.68
78–94%
BM25 (Okapi)
F1: 0.71
F1: 0.48
F1: 0.22
~45%
TF-IDF
F1: 0.68
F1: 0.44
F1: 0.19
~60%
Sliding window (fixed)
F1: 0.55
F1: 0.31
F1: 0.14
~50%
Heuristic importance scoring
F1: 0.62
F1: 0.41
F1: 0.18
~55%
LLMLingua (token compression)
F1: 0.58
F1: 0.35
F1: 0.16
~30%
Why BM25 & TF-IDF fail

Keyword frequency models. They find documents that mention the right words — not documents that contain the answer. On adversarial tests with decoy content, they collapse. No semantic understanding. No relevance scoring. No quality awareness.

Why sliding window fails

Assumes the answer is near the query in sequence. In real enterprise data — emails, records, filings — the needle can be anywhere. A fixed window misses it entirely on hard cases. No intelligence. Just position.

Why heuristics fail

Recency + length + keyword match. Works on simple cases. Fails when the needle is old, short, or uses different vocabulary. The QCE scores across multiple quantum mechanical dimensions simultaneously — heuristics score one dimension at a time.

Historical exploratory comparison retained for context; it is not part of the current authoritative evidence set and should not support a superiority claim. Current diligence should use the frozen six-domain and three-scale artifacts.

auto
Importance-signal evidence — a two-product story

Two test types.
One tells you exactly which selector fits.

We ran two distinct importance-needle scenarios across 500K, physical 5M and represented 500M corpora — using real multi-domain records, 20 planted targets at varying importance tiers, and adversarial hard negatives at maximum importance.

Test 1 — Importance-aligned

All 20 targets at importance = 1.0 (maximum). Hard negatives also at 1.0 but content actively denies the signal. When your metadata says it is critical — do we find it?

Test 2 — Importance-stratified

Same 20 targets but spread across tiers: 1.0, 0.75, 0.5 and 0.25. Hard negatives stay at 1.0. Do we find useful evidence even when it sits at a lower-priority tier?

Algorithm
Al 500K
Al 5M
Al 500M
St 500K
St 5M
St 500M
Latency
QCE importance-first (importance-first)
100%
100%
85–100%
30–35%
35%
45–50%
3–9 ms
QCE → QCE (query-aware)
100%
100%
100%
100%
100%
100%
500–5000 ms

Al = aligned · St = stratified. Each cell is one synthetic control case per scale. Latency: QCE at physical 500K–5M. QCE–QCE at physical 5M.

QCE — the speed champion

When upstream metadata is trustworthy, QCE finds every max-importance target in 3–9 ms at physical scales to 5M. Token savings remain above 98%. Ideal for critical-alert feeds, triage queues and any corpus where "importance" is genuinely signal-rich.

Query-aware — the universal choice

QCE through QCE returned 100% recall on both aligned and stratified at all three scales. They find evidence wherever it sits in the importance hierarchy — critical where relevance and priority metadata are not perfectly correlated.

A portfolio, not one algorithm

Customers with reliable importance metadata can deploy the ultra-fast importance-first family. Customers needing evidence at any priority level use the query-aware family. Routing can blend both in a single request.

auto
Future potential — beyond enterprise AI

The relevance engine
for everything that needs to find the right thing.

The QCE solves a universal problem: given a large set of items and a query, find the most relevant ones. This is not just an AI cost problem. It is the fundamental problem of every recommendation, discovery, and search system on the planet.

Vertical specialisation — now
Domain-tuned QCE

Each industry has different relevance signals. The current family spans v1–QCE; QCE adds query-aware lexical and corpus signals. Domain-specific weighting, semantic matching and held-out vertical validation are next-stage opportunities rather than completed claims.

Verticals: Healthcare · Legal · Financial · Defence · Nuclear · Government
AI intelligence layer — next
QCE + Domain LLM

The QCE selects the signal. A domain-trained LLM reasons over it. Together: a complete, private, auditable AI stack for regulated industries. No public LLM dependency. No data leaving the secure perimeter. Full-stack AI as a service.

Revenue: Platform fee + per-query + outcome-based
Beyond enterprise AI — future
Universal relevance

The same mathematics that finds the right medical record finds the right film, the right song, the right product. Netflix recommendation. Spotify discovery. E-commerce search. Any system that needs to find the most relevant item in a large set.

Market: Recommendation engines · Search · Discovery · Personalisation
The insight — why this generalises

Netflix has 15,000+ titles. A user query ("something like Inception but lighter") is a needle-in-haystack problem. The QCE scores every title across multiple relevance dimensions simultaneously — genre, tone, pacing, cast, viewer history — and returns the top candidates. The same quantum mechanical scoring framework. Different domain. Same mathematics.

Spotify has 100M+ tracks. "Something like this but more upbeat for a morning run" is a multi-dimensional relevance problem. BM25 and TF-IDF fail here — they're keyword models. The QCE scores across audio features, mood, tempo, listener context simultaneously. This is the same algorithm. Different data. Massive market.

18 / 28
Strategic value to acquirers

The AI cost war of 2025.
This is the weapon.

Every major AI platform faces the same challenge: enterprise customers are hitting cost walls. Whoever acquires Weavi-QCE gains an immediate, validated competitive weapon.

If Anthropic acquires
  • → Claude effective cost: $1.20 → $0.30/1M tokens
  • → Beat OpenAI and Google on price instantly
  • → Market share: 15% → 30%+ in 12 months
  • → 18-month competitive moat
If OpenAI acquires
  • → Prevent Anthropic gaining competitive weapon
  • → Protect $100B valuation and market share
  • → Justify premium pricing with superior economics
  • → Margin improvement: 35% → 60%
If Google acquires
  • → Illustrative licensing scenario: $612M/yr at 15% revenue share on Workspace Pro uplift
  • → "Only AI that searches your entire work history"
  • → 4× Workspace Pro adoption (McKinsey: AI features drive 3–5× premium tier conversion)
  • → Extend Gemini to "effectively unlimited" context window
The decision math
Build internally
$50M R&D · 18 months · competitor might acquire first
Acquire now
£75M–£125M · immediate advantage · validated technology in hand
Acquisition ROI diligence
Model avoided tokens × verified workload volume × gross-margin value; payback cannot be claimed before partner economics
19 / 28
Valuation — evidence-led framing

Technical risk reduced.
Commercial proof is the next value inflection.

The test programme strongly supports technical efficacy. It does not, by itself, establish product-market fit or a financing valuation. We separate a conventional venture valuation from the strategic value the technology could create for a large platform.

Evidence today
Validated

Working API, real-data testing, reproducible methodology and a broad algorithm portfolio reduce core technical risk.

Venture benchmark
£10M–£25M

Indicative pre-money range for a pre-revenue deep-tech financing; final pricing depends on team, patents, pilots and round terms.

Strategic outcome
£50M–£125M+

Possible only where independently reproduced savings create exceptional acquirer-specific value or competitive tension.

Value inflection
Pilots + IP

Patent filings, customer-side validation, paid pilots and credible LOIs are the fastest route to a materially stronger position.

CompanyAcquirerPriceContextYearvs Weavi-QCE
Inflection AIMicrosoft$650MTalent + IP licence. Team of ~70. Consumer AI product (Pi) with users. Structured as hiring + licence, not traditional M&A.2024Much larger team · consumer product · different deal structure
Adept AIAmazon$450MTalent acquisition. Enterprise automation demos, early pilots. Team of ~100.2024Larger team · earlier-stage validation
DeepMindGoogle£400M (~$650M)75 world-class researchers. Peer-reviewed publications. Demonstrated superhuman Atari performance.2014World-leading research team · published peer review
PerceptioApple~$200MOn-device AI inference. Small team. No revenue.2015Comparable team size · less statistical validation
Weavi-QCETBDNot yet pricedWorking API. Real-data validated. Reproducible results. Pre-revenue, pre-patent.2026Technical evidence strong · commercial evidence next
20 / 28
Exit scenarios

Three clear paths
to liquidity.

Path 1 — Strategic Acquisition
£75M–£200M · 90 days
  • → OpenAI: Prevent Anthropic gaining competitive weapon
  • → Anthropic: Make Claude effectively $0.30/1M tokens overnight
  • → Google: Extend Gemini to "effectively unlimited" context
  • → Databricks/Snowflake: Differentiate AI data platform
  • → Palantir: Defence intelligence fusion at scale
Path 2 — VC / Independent
£750M–£4B · 3 years
  • → Series A: £50M–£100M post-money valuation
  • → Capital needed: $15M–$30M
  • → Year 1 target: $120M ARR (platform partnerships)
  • → Year 2 target: $320M ARR (enterprise + scale)
  • → Year 3 exit: £1.5B–£4B valuation (with revenue proof)
Path 3 — IPO
£12B–£20B · 5 years
  • → Year 1: $120M ARR (platform partnerships)
  • → Year 2: $320M ARR (enterprise + scale)
  • → Year 3: $650M ARR (market penetration)
  • → Year 5: $1.2B ARR (industry standard)
  • → Public markets: "AI cost optimisation infrastructure"
What you get immediately on acquisition
Working technology
  • ✓ 16 registered variants plus a frozen QCE research candidate
  • ✓ Production API live at api.weavi.ai
  • ✓ Matched evidence across Google, OpenAI and Anthropic families
Validated proof
  • ✓ 96.5% recall at 500K→256K; 91.5% at physical 5M→1M
  • ✓ Real SEC EDGAR data — peer-reviewable
  • ✓ Security-focused architecture; formal assurance/compliance remains a milestone
Strategic assets
  • ✓ Patent-review-ready algorithm and execution IP
  • ✓ 21,130 recorded outcomes and accumulated failure knowledge
  • ✓ Measured token economics; energy/carbon telemetry is a future diligence item
auto
Planet — the AI energy and water emergency

AI is consuming the planet.
Every token saved is a real resource saved.

These are not projections. These are verified 2024–2026 published findings from the IEA, Goldman Sachs, Microsoft, Google and the EPA. The scale is staggering — and it is accelerating.

⚡ Energy — The IEA Warning
100+ TWh
AI data centre electricity in 2024 alone
  • → Projected 200–400 TWh by 2027
  • → One AI query uses 10× the energy of a Google search
  • → Google: 3.2 quadrillion tokens/month and growing 7× year-on-year

IEA Electricity 2024 Report; IEA AI & Energy Supplement 2024; Google I/O keynote May 2026.

💧 Water — Goldman Sachs Finding
700,000 L
Fresh water to train GPT-3 once
  • → 1.8 L of cooling water per kWh of compute
  • → Microsoft: 6.4 billion gallons in 2022, +34% YoY from AI
  • → Google: 5.6 billion gallons in 2022 (their own disclosure)

Goldman Sachs "AI's Growing Water Footprint" 2025; Microsoft Sustainability Report 2024; Google Environmental Report 2023.

🌍 Carbon — The Trend
1.7%
Projected AI share of global electricity by 2026
  • → Equivalent to adding a country the size of Germany to the grid
  • → 0.233 kg CO₂/kWh UK grid; 0.386 kg CO₂/kWh US average
  • → EU CSRD mandates corporate AI emissions disclosure 2024+

IEA 2024; DESNZ 2024 grid intensity; EPA eGRID 2023; EU CSRD regulation 2023/2775.

Weavi QCE — the proportional reduction
Every token not processed…

…is compute not run → energy not consumed → water not evaporated → carbon not emitted. QCE's token reduction maps directly to proportional compute avoidance on inference infrastructure.

Illustrative: Google at QCE scale

3.2 × 10¹⁵ tokens/month × 50% QCE median reduction × 0.001 kWh/1K tokens (IEA 2024 inference estimate) = ~1.6 TWh saved per month at one company. Proportional water and carbon follow.

The important caveat

Energy reduction requires measured deployment telemetry (actual compute, cooling PUE, grid mix). The token saving is measured; the proportional resource saving requires a partner pilot with instrumented infrastructure to confirm.

auto
Planet — QCE as green AI infrastructure

The AI layer that
does more with less.

At 50–80% typical input reduction, the proportional resource impact at enterprise scale is material. Below are illustrative estimates from published energy, water and carbon intensities — pending verified deployment telemetry.

1–3 TWh
estimated energy per enterprise per year

At a large bank running 1B tokens/day with QCE 80% reduction and IEA's inference estimate of 0.001 kWh/1K tokens.

IEA AI Energy Supplement 2024 · pending deployment telemetry to confirm.

Billions of litres
cooling water per major platform per year

1.8L/kWh cooling ratio × energy saved from token reduction = proportional water avoidance.

Goldman Sachs AI Water Footprint 2025 · IEA cooling intensity estimate.

Board-level ESG
reportable AI efficiency metric

Token reduction per query is the auditable input metric. With deployment telemetry, it chains to energy and carbon for EU CSRD, UK TCFD and US SEC climate disclosures.

EU CSRD 2023/2775 · UK TCFD mandatory 2026 · SEC Climate Rule 2024.

International regulatory context — sustainability mandates now apply to AI
EU — CSRD
50,000+ large companies must report AI infrastructure emissions from 2024. QCE creates the auditable efficiency metric.
UK — TCFD
Mandatory climate disclosure for listed companies. AI compute increasingly classified as Scope 2 emissions.
US — SEC Climate Rule
Public companies disclose material climate risks including data centre footprint. AI inference is in scope.
APAC — Singapore, Japan
MAS Net Zero commitment; Japan Carbon Neutrality 2050 — both include data centre AI efficiency targets.
The strategic framing

QCE is not an ESG story with a technology attachment. It is a technology that inherently improves ESG metrics by solving the root problem: AI processes far more data than it needs. The resource savings are a direct consequence of the economic efficiency — not an overlay or offset scheme. That makes them genuine, auditable and repeatable.

ESG 1
ESG impact — the carbon story

Fewer model tokens.
A measurable efficiency opportunity.

Token and provider-cost reductions are measured. Energy and carbon are not assumed proportional: hardware, batching, caching, utilisation and grid mix require customer/provider telemetry.

53.17%
median paired model-token reduction

Latest matched 500K Sonnet/Gemini/Opus calls. Pair counts 6–8.

80.0%
declared input reduction at physical 5M

With 91.5% labelled algorithm recall in the latest matched diagnostic.

Pilot metric
energy and carbon per successful query

Requires measured provider/customer compute and matched-quality completion.

Auditable next step

A customer pilot should record successful-query quality, model/provider tokens, QCE compute, accelerator utilisation, energy and grid intensity. Only then should avoided-energy, carbon or CSRD reporting claims be calculated and independently reviewed.

ESG 2
ESG impact — the regulatory & commercial case

Carbon reduction
is now a board-level mandate.

Sustainability reporting increases demand for measured infrastructure efficiency. Weavi-QCE already records token economics; reportable energy or carbon reduction requires deployment telemetry, an agreed boundary and independent review.

The regulatory driver
  • EU CSRD (2024): Mandatory sustainability reporting for 50,000+ large EU companies
  • UK TCFD: Mandatory climate-related financial disclosures for listed companies
  • SEC Climate Rule (2024): US public companies must disclose material climate risks
  • → AI compute is now a material, reportable emissions source for any large enterprise
  • → Weavi-QCE records token reduction; energy and avoided-carbon claims require measured infrastructure telemetry
At scale — the numbers
To measure
CO₂ per successful query
To verify
provider energy reduction

Token savings do not automatically scale linearly into energy or carbon savings. A customer/provider pilot must establish the conversion and reporting boundary.

Green technology incentives
  • → UK: R&D tax credits + Green Investment Bank eligibility
  • → EU: Green Deal taxonomy — sustainable technology classification
  • → US: IRA clean technology credits applicable to AI efficiency
  • → ESG fund eligibility — growing pool of sustainability-mandated capital
auto
The multi-source advantage

One claim. Five data sources.
One precise answer.

Think of a large insurance claim. Multiple organisations hold relevant data. Weavi-QCE ingests all of it, finds the relevant signal across every source, and delivers only that to the LLM.

Multi-source ingestion
Insurance claim — major incident
Police report — 25K tokens
Medical records — 50K tokens
Witness statements — 15K tokens
Underwriter data — 30K tokens
Policy documents — 20K tokens
Total: 140,000 tokens across 5 sources
Weavi-QCE extracts relevant signal → ~8,000 tokens (~94% reduction in this scenario)
~94%
Token reduction in this scenario

Across all 5 sources combined — results vary by dataset

Material
Fraud detection improvement

Cross-source signal quality improves significantly

Significant
Adjuster productivity gain

More claims reviewed per day with better context

Any org
Multi-source ready

Gov, mil, edu, health, legal, finance

The key insight

The LLM is not a database index. Weavi-QCE performs final question-aware evidence selection across the authorised candidate records it receives, so the LLM can focus on reasoning. Enterprise-wide discovery remains a connector/indexing layer.

auto
Current status & traction

Where we are today.
Honest. No spin.

What is built and working
  • ✓ Production API live at api.weavi.ai
  • ✓ 16 registered variants (v1–QCE), plus an unregistered QCE research candidate
  • ✓ Durable Evidence Lab history with real model, selector and combined-pipeline outcomes
  • ✓ 2,574 recorded model/planner requests across Google, OpenAI and Anthropic workstreams
  • ✓ Real SEC EDGAR integration — live data, zero synthetic
  • ✓ FDA adverse event reference application — working demo
  • ✓ Complete technical documentation suite
  • ✓ Broad retrieval and statistical diagnostics, with reference-validation roadmap defined
What we are building toward
  • → Provisional patent filings (immediate priority)
  • → Independent benchmark package and partner-side replication
  • → Design-partner and paid-pilot conversations
  • → NIST / ISO / SOC 2 Type II certification (post-funding)
What we don't have yet — honest
  • ✗ Revenue ($0 — pre-revenue stage)
  • ✗ Paying customers (0 — technology validation phase)
  • ✗ Filed patents (provisional filing is immediate next step)
  • ✗ Sales team (funded by Series A or acquisition)
The comparable context

Inflection AI and Adept AI were primarily talent and IP transactions involving much larger teams and materially different circumstances. They illustrate strategic appetite for AI capability, but they are not reliable valuation comparables for Weavi-QCE. Our value must be established through IP diligence, customer-side validation and partner-specific economics.

24 / 28
Commercial milestones

Evidence before forecast.
Build the model from paid usage.

At pre-revenue stage, precise five-year ARR and valuation claims create false confidence. The operating plan is milestone-based; a forecast follows once pricing, sales cycle and expansion behaviour are observed.

PeriodCommercial proofProduct proofDecision unlocked
0–3 months3 design partnersPartner-owned corpus replicationBeachhead and success criteria
3–6 monthsFirst paid pilotProduction reliability and security scopeInitial pricing and deployment model
6–12 months3–5 paid deploymentsRepeatability across customersBottom-up forecast and seed/Series A plan
12–24 monthsExpansion revenueVertical tuning and platform integrationScale direct sales or channel distribution
Economics to measure

Compute cost, support load, realised customer saving, gross margin and willingness to pay will be measured in pilots rather than assumed.

Capital priorities

Customer engineering, production hardening, patents, security assurance and focused enterprise development — sequenced against evidence.

Forecast trigger

Publish a base, upside and downside model after the first paid deployments establish price, usage, sales cycle and retention inputs.

25 / 28
IP & potential moat

What could become
genuinely defensible.

The algorithm portfolio

Sixteen registered variants from v1 through QCE cover different budget and selection behaviours. The unregistered QCE research candidate leads the latest 500K and physical-5M semantic tests. The moat is measured quality, execution evidence and accumulated failure knowledge—not variant count.

Patent strategy
  • → Novel multi-dimensional scoring methodology
  • → Quantum-inspired information selection process
  • → Auto-routing algorithm selection system
  • → Prior-art review followed by focused provisional filings where counsel confirms a defensible claim
  • → Preserve trade-secret implementation details alongside any patent strategy
Accumulated implementation lead

The portfolio reflects years of framework development and multiple algorithm variants. The duration and commercial value of that lead must be established through prior-art work, strong comparator benchmarks and deployment learning.

Comparator programme
MethodStatusPurpose
Full-context LLMImplementedCurrent end-to-end baseline where context permits
Importance controlsImplementedAligned and stratified importance-needle controls
BM25 / TF-IDFPlannedClassical sparse-retrieval controls
Dense / hybrid RAGPlannedModern semantic production baseline
QCE ablationsImplementedPreserve accepted and rejected candidate hypotheses
Deployment learning

With appropriate customer permission and privacy controls, deployment telemetry can improve routing and domain configuration. That learning loop is a potential future moat—not current traction.

26 / 28
Compliance & security

Enterprise-grade
from day one.

Regulated industries — healthcare, finance, defence, government — require compliance before they can deploy. Weavi-QCE is designed for this from the ground up.

Content-minimising architecture

The bounded API is designed not to persist or content-log synchronous customer records. Data is transmitted to the selected deployment endpoint; customer-controlled/VPC execution is a future deployment option requiring validation.

Healthcare compliance path

Data minimisation and bounded processing support future healthcare deployment. HIPAA applicability, BAAs, risk analysis, operational controls and customer validation are required before any compliant-production claim.

GDPR-oriented controls

Data-minimising processing can support purpose limitation. Controller/processor roles, lawful basis, DPAs, residency, deletion and deployment-specific assessment remain necessary.

SOC 2 roadmap

Security, availability and confidentiality controls are being built toward an assurance scope. No SOC 2 Type II report exists yet; readiness and audit timing require an independent assessor.

EU CSRD (ESG reporting)

Token reduction is measurable, but energy and carbon are not necessarily proportional because provider hardware, batching, caching, utilisation and grid mix vary. Customer/provider telemetry is required before any CSRD or avoided-emissions claim.

Basel III / Solvency II

The evidence suite includes paired tests, intervals and multiplicity correction relevant to model-risk work. It has not been approved by a regulator or demonstrated as sufficient for Basel III/Solvency II compliance.

On-premise deployment

For classified environments (defence, nuclear, government): algorithm runs on customer's own infrastructure. No cloud dependency. No data leaves the secure perimeter. TS/SCI clearance pathway available.

27 / 28
The ask — prove it together

Run the QCE on your data.
Measure the result before discussing price.

The strongest next proof is customer-side validation. We are seeking strategic partners, design customers and investors who can help turn reproducible technical evidence into commercial evidence.

Strategic validation
  • → Bring a representative corpus, target query set and preferred LLM
  • → Agree success criteria before the test: recall, precision, cost and latency
  • → Compare baseline and QCE-assisted results under identical conditions
  • → Review the methodology and outputs with your own technical team
Capital and strategic path
  • → Open to strategic investment, licensing, a venture round or acquisition
  • → Capital funds customer pilots, production hardening, patents and security certification
  • → Structure and valuation should reflect diligence, pilot scope and partner-specific value
What you get immediately
  • ✓ 16 registered algorithm variants plus the latest SEIF research candidate
  • ✓ Production-ready API infrastructure (live at api.weavi.ai)
  • ✓ Matched evidence across Google, OpenAI and Anthropic model families
  • ✓ Complete test methodology (peer-reviewable)
  • ✓ Zero-retention processing architecture
  • ✓ A differentiated, patent-review-ready algorithm portfolio
Next value milestones
  • → Independent replication on a partner-owned dataset
  • → Provisional patent filings following professional prior-art review
  • → Three design partners and the first paid production pilot
  • → Measured unit economics, reliability and deployment requirements
27 / 29
Risk factors — honest assessment

What could go wrong.
And why we're prepared.

Investors respect honesty. Here are the real risks — and our mitigation for each.

Risk: AI providers build this themselves

Likelihood: Medium. OpenAI/Anthropic could attempt to build context optimisation. Mitigation: 18–24 month technical lead. Our validation methodology is peer-reviewable — theirs would not be. First-mover enterprise contracts create switching costs. And: if they're building it, they should buy it instead.

Risk: LLM context windows keep expanding

Likelihood: Certain — but irrelevant. Even Gemini's 2M token window covers only 0.6% of a typical enterprise user's data. Google processes 3.2 quadrillion tokens/month — and still wastes 78% of them. Larger windows = larger waste = larger savings from Weavi-QCE.

Risk: No revenue, no customers yet

Likelihood: Current reality. We have £0 revenue and 0 paying customers. Mitigation: The API is live and the test results are reproducible and reviewable. The next milestone is independent customer-side validation; financing and strategic value should be discussed in that context.

Risk: Enterprise sales cycles are long

Likelihood: High for VC path. Enterprise deals take 6–18 months. Mitigation: Platform partnerships (Databricks, Anthropic) bypass direct enterprise sales — one deal unlocks thousands of customers. Acquisition path avoids this entirely.

Risk: IP not yet patented

Likelihood: Current reality. No patents filed yet. Mitigation: Provisional patents can be filed within 30 days for approximately £8–25K. The algorithms are novel and patent-worthy. Trade secret protection applies in the interim. Filing immediately materially strengthens the IP position and valuation.

Risk: Small team

Likelihood: Current reality. Early-stage team. Mitigation: The technology is built and proven. Series A funding adds enterprise sales, legal, and engineering. Acquisition path integrates into acquirer's team immediately. Perceptio (acquired by Apple, ~$200M, 2015) was a similarly small team with novel IP and no revenue.

28 / 29
Appendix — full test results summary

Current evidence ladder.
Quality and savings reported together.

15 Sep 2026 Claude Sonnet 5 · Gemini 3.8 Flash · Claude Opus 5 6 public domains · real SEC · OpenFDA

Every result has a create-once hash-locked artifact. Scale-specific caveats and negative outcomes remain part of diligence.

Test scenarionReductionRecallMAPMRR
QCE · physical 500K→256K1048.8%96.5%.5387.6880
QCE · physical 5M→1M1080.0%91.5%.3996.5599
QCE · represented 500M→5M1099.0%62.5%.3918.5576
QCE · six-domain locked validation21052.9%87.1%.6952.8069
Gemini-facet QCE · discovery32851.9%89.4%.7263.8341
GPT-facet QCE · discovery32851.9%90.0%.7053.8020
80.01%
Median reduction · latest 30 semantic scale cases
898
Independently labelled query units
21,130
Recorded arm-level outcomes

Arm-level outcomes include ablations and scale points; they are not independent samples. Precision uses incomplete public qrels. Partner-owned labels, strong external retrieval comparators and independent replication remain decisive next standards.

29 / 29
"The investment case is not that every token can be removed. It is that the right evidence may be preserved with far fewer of them."

Weavi-QCE is a working, pre-revenue infrastructure asset with differentiated IP, a live API and promising internal evidence. The opportunity is meaningful; the remaining work—independent benchmarks, patents and customer proof—is explicit and measurable.

We are looking for a thoughtful partner who can bring a representative corpus, help define success criteria and evaluate the technology seriously. Strategic investment, licensing, a venture round or acquisition can follow the evidence rather than precede it.

Live API
Working infrastructure
Real data
SEC + OpenFDA routes
3 families
Google · OpenAI · Anthropic evidence
Pre-rev
Commercial proof next

© 2026 Weavi. Commercial in Confidence. E & OE. Internal benchmark observations are subject to diligence and independent replication. This document does not constitute a public offer of securities.

1 / 28