A real agent, or RPA in a new coat
A forecast reorganised the year: more than 40% of agentic projects will be cancelled by the end of 2027 — for runaway cost, diffuse value, or inadequate risk control — and the aggravating factor has a name, "agent washing", rebranding an assistant, a chatbot or an RPA bot as an "agent" with no real autonomy. For the board, what separates the project that survives from the write-off isn't the demo — it's measurable autonomy, ROI attributed per flow, and risk control proportional to blast radius. The autonomy test that breaks the happy path on purpose, ROI per flow as the continuity gate, controls sized to what the agent can touch, the total cost that includes the reasoning loop, the four ways a board ends up in the 40%, and why the document desk is the cleanest place to run the test.
13 minBuild the moat, partner for the plumbing
One figure from the last year cut through a lot of internal pride: programs run with a specialised partner reach production at roughly twice the rate of those built purely in-house — about two thirds versus one third. The naïve reading is "outsource it", and that's wrong; so is "we build everything". The build-vs-partner call decided layer by layer — rent the model, partner for orchestration and evaluation, own the domain data and the corrected flow — the portability clause that keeps a head-start from becoming lock-in, who should own the customer's correction, the four ways a partner program quietly fails, the honest trade-off between speed now and margin later, and why the board metric that predicts who ships isn't ML headcount.
13 minThe gap isn't the model. It's the instrumentation
Roughly 95% of GenAI pilots deliver no measurable P&L impact, almost everyone has a pilot and almost nobody has a fleet in production. The cause isn't model quality and it isn't regulation — it's the three layers underneath the technology that never got built: measurement that proves the task works, integration that wires it into the flow that creates value, and ownership that keeps the system learning after go-live. The three layers in detail, the build-vs-partner signal that roughly doubles the production rate, the four ways the evaluation layer quietly rots, the honest trade-off of instrumenting before you see a number, and why "which model" is the wrong opening question.
13 minTemplates or zero-shot, decided by the tail
Two honest ways to get fields out of a document, and most of the noise comes from pretending one has to lose. Template-based extraction is deterministic, fast and near-free per page — and breaks the instant a layout drifts. Zero-shot absorbs the long tail of formats you'll never template, at a per-call cost and with quiet, plausible failures. The eight dimensions a buyer actually weighs, where each genuinely wins, the hybrid that ships — route by class and confidence, zero-shot as cold-start and fallback, templates as a cache of what zero-shot learned — and the failure mode that bites each when nobody's watching.
12 minUpskilling at the speed of automation
In 12 months agentic-AI upskilling went from an optional internal program to a board investment — consultancies certifying tens of thousands of professionals in a single deal, banks certifying whole operations functions. The dual thesis (operating agentic AI needs a different skill stack; the firm that certifies at scale first captures the talent supply and builds a switching cost), the certification-as-moat case and exactly where it rots into paper, a competence matrix by role that survives an audit, the four ways an upskilling program quietly fails, and why the document desk is the cleanest place to make the certificate predict the work.
13 minIntelligence per worker, measured honestly
Frontier firms run ~3.5x more intelligence per worker than the median, and in 2026 the metric jumped from HR slide to board agenda — arriving in the same news cycle as the AI-restructuring layoffs. It is a real measurement and an easy euphemism, one keystroke apart. The four components that make it auditable instead of anecdotal — output per FTE-equivalent, delegable-task automation rate, time recovered with a destination, quality with vs without the agent — why measuring only the numerator books six months of gain against a year of cost, and where the document desk measures it honestly first.
13 minAlways-on agents and the back-office that stops opening tickets
2026 shipped a category that 2025 treated as roadmap: agents that run continuously, watch a queue, decide and execute without a human trigger. The five back-office workflows that absorbed the pattern first, the four design patterns that consolidated — subscription / idle, per-agent budget, kill switch and dead-man's switch, ground-truth-of-when-to-stop — and the cost trade-off that survives a CFO's diligence.
13 minCapability contracts for the post-app era
The 2026 device-and-OS shift moved the buyer one layer up: the assistant in the operating system, the notebook and the phone decides which portal to call. The three consequences for SaaS in the next two quarters, the four-artefact capability contract buyers are asking for by name — capability manifest, intent registry, tool catalog with permission scope, per-agent telemetry — and the anti-patterns that quietly retire a portal.
12 minRegulator-in-the-loop: pre-launch model access as the 2026 baseline
Frontier-model providers in the US, UK and EU now hand pre-release weights to government safety institutes before public availability. The five RFP questions enterprise procurement put on the table in 2026, the model-card fields buyers check, the rollback architecture that holds when a regulator pulls a version — and the three anti-patterns that disqualify in regulated diligence.
13 minWhen inference owns your margin: compute as the 2026 moat for document AI
Record-breaking AI capital — an US$ 122B round at US$ 852B post-money, US$ 200B+ in committed compute, an OCR floor sliding to US$ 2/1,000 pages. The three decisions a CIO and a CFO have to make now, the five-line cost build-up procurement is asking vendors to show, and the two vendor profiles most likely to get squeezed in the next 12–18 months.
12 minWhen reasoning carries a proof: verifiable frontier models for enterprise documents
2026 moved the frontier-model line from "better answers" to "checkable steps". The four enterprise places verifiable reasoning already landed — audit, compliance, legal, regulated extraction — the five-component architecture that ships, the 78–92% auditor-accepted rates in early deployments, and the limits we are not going to pretend away.
12 minAgent-as-user: the post-app B2B portal
A measurable share of B2B portal sessions in 2026 originate from an assistant acting for a logged-in buyer. Six surfaces an agent actually needs, four moves in the playbook that ships, and the anti-patterns that quietly disqualify a portal — read against what changes for the document AI behind it.
11 minSovereign AI: residency as a contract clause, not a policy slide
Sovereign AI consolidated as an enterprise category in 2026, with a 30–60% premium and six clauses regulated buyers now want by name — reserved compute, jurisdiction-bound models, BYOK/BYOM, signed audit trail, kill switch and bordered sub-processors. The architecture that holds and the anti-patterns that fail diligence.
12 minAI operating model: six artifacts, three anti-patterns, 90 days
76% of large orgs hired a CAIO, but the title alone does not survive an RFP. The six artifacts buyers now ask for, the three anti-patterns we keep seeing, and a 90-day build plan that fits inside one quarter.
12 minAI governance for document AI: ISO 42001, EU AI Act and audit evidence in 2026
Gartner expects 70%+ of enterprises to adopt a formal AI governance standard by end of 2026, and the platforms market goes from US$ 492M to US$ 1B by 2030. The five frameworks, three audit shapes and the evidence envelope that holds.
12 minGenerative document fraud: why detection became a board item in 2026
AI-generated document fraud grew nearly 5x in eight months, and 97.8% of risk leaders say it already keeps them up at night. The four shapes hitting the queue, the hybrid stack that holds, and what an investigation agent actually does.
12 minMulti-agent systems in the backoffice: practical orchestration in 2026
One agent stops scaling at the third edge case. Squads of specialists coordinated by a maestro hit 35–55% more throughput on complex flows. The three patterns, the five failure modes, and the observability shape that ships.
12 minVLMs and the document pipeline collapse: when one model replaces four stages
Vision Language Models fold OCR, classification, layout analysis and extraction into a single call. When the collapse pays back, where the classical stack still wins, and the hybrid routing most production programs actually ship.
11 minFrom reactive to predictive: when document AI starts forecasting the work
The next IDP shift isn't faster extraction — it's pipelines that forecast backlog, flag the contract clauses legal will redline, and order the queue by expected value. Three workflows rewritten and the parts that quietly break.
11 minFrom extraction to execution: IDP as the engine, not a step
Gartner expects 40% of enterprise apps to ship task-specialized agents by end of 2026. What actually changes when IDP stops feeding the workflow and starts running it — three workflows rewritten and the parts that quietly break.
11 minReasoning models: when document AI thinks before it extracts
Reasoning models cost 5–10x more per call. They cut human review 50–70% on the right class of documents. What pays back, what doesn't, and the routing pattern that decides whether the program ships.
11 minFrom RPA to agents: the autonomy bar 2026 is asking documents to clear
Gartner expects 30% of enterprises to have automated more than half their workflows by end of 2026, up from under 10% in 2023. The architectural difference between RPA and agents — and the four failure modes that come with the upgrade.
11 minThe AI ROI gap: what the 29% do differently
Enterprise AI budgets jumped 65% to US$11.6M, but only 29% report meaningful ROI and 42% have abandoned most of their initiatives. Four traits separate programs that pay back from those that quietly stalled.
10 minOCR in 2026: from 70% to 98% — and why the pipeline didn't disappear
Multimodal LLMs cleared 98% on printed text and 95% on handwriting. The interesting question stopped being whether to adopt them — it became where in the pipeline they actually pay back.
9 minFrom extraction to decision: how IDP stopped copying fields and started thinking
67% of enterprise document programs are evaluating agentic IDP, up from 23% two years ago. What changed in the reference architecture, and the parts that are quietly harder than the slide suggests.
10 minPrompt injection in document AI: the threat model nobody scopes
Every document your pipeline ingests is untrusted instruction text. The threat model, three real attack patterns, and the four defenses that actually hold.
9 minSOC 2 Type II for AI startups: what to build in
The five architectural commitments that turn the SOC 2 audit from a quarter-long cleanup project into an emergent property of your platform.
9 minSub-processors in AI: what your DPA needs in 2026
AI products invoke 4–7 sub-processors per request. What your DPA needs to say about LLM providers, observability vendors, and zero-retention APIs in 2026.
9 minGDPR for document AI: a practical guide for operators
Article 28 obligations, lawful basis, sub-processor governance, data subject rights, and what your DPA actually needs to say.
9 minTracing agentic document extraction
How to make multi-step LLM workflows debuggable. OpenTelemetry span design, sampling strategies, and the structured logs that turn a black box into a flight recorder.
8 minAudit trails for non-deterministic outputs
How to log AI extractions in a way that holds up to reproducibility, regulatory audit, and customer "why did you extract this?" questions — with the actual schema we use at Cogneris.
8 minFrom 48h to 2.4s: a fintech underwriting case
How a fast-growing fintech replaced a team of 12 operators with an API.
5 minPrior auth turnaround cut from 5 days to 4 hours
How a regional health plan automated PA triage with HIPAA-grade controls.
6 minContract intake at 100x volume — same legal team
How a US M&A boutique scaled diligence with Cogneris's extraction API.
5 min