What is Cognibl? Work management for people and agents

Most enterprises now use AI somewhere, yet very few capture material value from it. McKinsey's 2025 State of AI survey found that around 88% of organisations use AI in at least one function, while only about 6% are what it calls AI-mature, the small group capturing material EBIT impact. Stanford's 2026 AI Index reports that fewer than 10% of enterprises have fully scaled AI in any function. One reason the gap stays so wide is quiet and specific: when an AI agent does a real piece of work, there is often no dependable way to tell whether it is genuinely finished or merely reported as finished.

Cognibl, from DILR.AI, is a work-management platform built for that exact problem. It is a tracker for teams where people and AI agents work side by side, and it will not let a task reach a done status until proof of the work is attached. Before going further, a word on the name, because it is easily confused with Cognism, an established and entirely separate company. The two names sit one letter apart yet the products do not overlap, which is the first thing this guide settles.

This article answers the plain question, what is Cognibl, for an operations lead, a head of delivery or an engineering manager deciding whether to let AI agents into a team's workflow. It sets out what the product is, what it is not, who it is for and where to verify every claim. It does not re-run the deeper operating-model argument. For how a proof-of-done gate reshapes an entire delivery process, our AI agent work management guide carries that in full, and this page stays at the level of a first, honest introduction.

This guide is shipped by the team behind Cognibl, work management for teams where people and AI agents share one board. Or see DATS, our senior-led AI consulting and delivery system for organisations placing AI into production.

What is Cognibl?

Cognibl, from DILR.AI, is a work-management platform for teams where people and AI agents share one board. Agents pick up tasks under their own name against the same statuses the team uses. What sets Cognibl apart is a done-gate: a task reaches a done status only once a proof of work is attached, and the database refuses the move without one. The same rule holds for a person and for an agent.

The idea is best understood as a trust layer sitting between the agent and the business, so work is finished when it is proved rather than when it is claimed. On an ordinary tracker, moving a task to done is a dropdown, and the claim is the only evidence. Cognibl replaces that claim with a record: the definition of done, the evidence behind it and every tool call are attached to the task rather than buried in a thread nobody can find later. The product describes itself as work management with an AI harness, so the project explains itself, and that phrase is a fair summary of the whole design.

Cognibl or Cognism: which were you looking for?

If you searched for Cognibl and kept getting Cognism, you were looking at two different companies. Cognism describes itself as a leader in premium sales intelligence, and it serves a different buyer entirely. Cognibl, from DILR.AI, is a work-management and delivery tool for teams running work through AI agents. The names sit one letter apart, but the products do not overlap.

This matters because Cognibl is a new name, and a new name has little search presence yet. It is easy to land on the older, established term by mistake, which is ordinary for any product early in its life and says nothing about which tool fits your need. The practical guidance is simple: if you want contact data and prospecting, Cognism is the sales-intelligence company you mean; if you want a board where people and AI agents share the same work and prove it is done, Cognibl is the one built by DILR.AI. The rest of this guide is about the second.

How is Cognibl different from an ordinary project tracker?

Cognibl is a project tracker first, with a backlog, a sprint board, a roadmap and version-control links, built for keyboard-first speed. The difference is what happens at the edges of a task. On a classic tool, completion is a status someone selects, and audit is whatever an integration happened to log. In Cognibl, completion is gated on attached proof, and what an agent may touch is set by a policy map rather than an API token.

Those are structural differences, not features bolted on, which is why they change how a team works rather than how a dashboard looks. Cognibl is also not a framework for building AI agents, such as CrewAI or LangGraph; it is the work board where a team tracks and proves work, whoever or whatever does it. This guide keeps the comparison short on purpose, because the full head-to-head against a classic issue tracker, capability by capability, belongs in the proof-of-done guide rather than a definitional page. The short version is that Cognibl is the tool a team could adopt purely as a tracker and still use every day, with the agent-governance and proof machinery available underneath when it is needed.

What is a proof of work in Cognibl, and why can a task not reach done without one?

In Cognibl, a proof of work is the evidence a task carries when it claims to be complete: a CSV describing the run, with the artefact, screenshots and hashes referenced inside it, plus a coverage report where tests are part of the claim. In Cognibl a status change to done is refused unless that proof is attached, and retrying does not change the answer. The proof is then read by whoever reviews the work, a person or a reviewing agent.

The gate exists because a claim is cheap and evidence is not. Anyone, human or agent, can say a task is finished; far fewer can attach a record that shows it. By moving the burden onto the proof rather than the assertion, Cognibl makes a reviewer's job a reading task instead of an act of faith. The flow below shows the path a single task takes, and it is deliberately the same for a person and for an agent.

The record left behind is written by the gateway, append-only and hash-chained. Editing a library item mints the next version rather than overwriting the last, identical content is refused, and nothing is deleted, so a trace from six months ago still resolves to the exact instructions that ran. That permanence is what turns a work board into an audit surface, which is where the next question leads.

How does Cognibl govern what AI agents are allowed to do?

Cognibl governs agents through an MCP gateway that is deny by default. Agents list and fetch the skills and definitions they need through it, but the switch is off until a person turns it on, so a toolset that has not been enabled is refused rather than silently missing. Every call is traced and every refusal is mirrored to audit. Two AI flows, one that validates proof and one that writes the project status, summarise and flag but never decide.

That design reflects where regulation is heading. The EU AI Act, in Article 14, requires that "High-risk AI systems shall be designed and developed in such a way, including with appropriate human-machine interface tools, that they can be effectively overseen by natural persons during the period in which they are in use." That duty binds the providers of high-risk systems rather than every work-management tool, and Cognibl is not making a compliance claim here. The principle, though, that a person can always oversee and read back what a machine did, is exactly the assumption Cognibl is built on, and it is the same instinct behind the ICO's guidance on AI and data protection, which any organisation should read before adopting an agent-capable tool. For teams that want the governance decided before the tooling, our AI operating model service sets the RACI and lifecycle first.

What Cognibl measures

Cognibl reports five delivery metrics that a business owner and an engineering lead can read from the same screen. Median cycle time is split into specifying, building, verifying and settling, so a team can see where time actually goes. First-pass verification rate shows the share of agent tasks that pass on the first attempt. Human wait share measures how much of a cycle was spent waiting on a person. Verified throughput counts completions that cleared the gate, and reopen rate tracks tasks reopened within thirty days.

The choice of metrics is itself a statement. Raw task counts and signup totals are deliberately absent, because a number that only ever rises measures effort rather than delivery. These are the engineering-delivery signals that describe how work actually flows, set alongside the classic figures a business owner already reports on, so nobody has to translate between two dialects. The deeper point, that specification quality shows up as a low first-pass rate long before it shows up in production, is the through-line our AI developer productivity work is built around.

Who is Cognibl for?

Cognibl is for teams that have started running work through AI agents and now need to trust the output the way they trust a person's. In practice that means an operations manager, a head of delivery, an engineering lead or a COO in an organisation where agents are moving from experiment to everyday work. It suits software delivery most obviously, and it can fit other functions where work is specified, done, verified and reopened.

The common thread across those buyers is accountability rather than novelty. A team adopts Cognibl when the question has shifted from can an agent do this to how do we know it did, and who will stand behind the result. That is a delivery-governance need, not a chatbot need, which is why the tool reads as a tracker with an audit spine rather than an AI assistant. For organisations that want the placement decided first, where AI genuinely belongs in a workflow, our AI placement diagnostic answers that before any tool is chosen.

How is Cognibl priced, and what do the tiers include?

Cognibl publishes three tiers by name, Free, Business and AI. Its stated position is that the free tier is the whole product, the cap being how many people can be in the team rather than which features they use. The Business tier lifts that member cap on the same product, and the AI tier adds two optional flows that read the proof and write the project status. This guide names the tiers rather than their fees.

That structure is worth understanding on its own terms. The Free tier carries the full tracker, the proof gate, the versioned skills library and the audit trail, with team size the only limit. The paid tiers scale with headcount, and the AI tier is where the two optional summary flows sit. To confirm exactly what each tier includes today, read the live plans at cognibl.com, since that page is the authority and this one is not.

Is Cognibl one of the other DILR products, like Dilr Voice?

No. Cognibl is one line in a portfolio built by DILR.AI, and it is distinct from the others. The voice platform answers phone calls, the clinical small language models read documents, the tutor diagnoses then teaches, and the content tool generates assets. Cognibl is the work-management line, the board where people and agents share and prove their work.

In product terms, Dilr Voice handles enterprise phone calls, Dilr Mira is a set of private clinical models, Dilr Academy is the AI tutor and DILR Studio generates content, while Cognibl tracks and proves the work. What ties the portfolio together is one operating philosophy rather than a shared feature set: place AI where the work and the value actually are, and leave a record a person can inspect. If you arrived here comparing DILR products, this is the one about how work is tracked and proved, not how calls are answered or documents are read.

Who builds and runs Cognibl?

Cognibl is built and run by Dilr.ai Ltd, a company registered in England in 2025, Company Number 16842656, based in Harrow, England. The same team ships DILR.AI's other products and advises organisations on where AI creates measurable value. Cognibl is the work-management line of that portfolio, so support and roadmap decisions sit with a single UK entity.

That wider advisory practice is delivered through the DATS consulting system, which places AI into production for organisations that want senior-led delivery around a tool like this. You can talk to the team directly to check fit.

Where can I verify what Cognibl does?

Every product claim on this page can be checked at source. The live product, its current plans and its documentation sit at cognibl.com, and DILR.AI publishes its own overview of the product as well. For the deeper argument about why a proof-of-done gate matters for agent work, this blog carries a full enterprise consulting guide and related engineering field notes. The links to all of them sit in the paragraph below.

Open the live product at cognibl.com, read the enterprise AI consulting guide, the field notes on evaluation harness engineering, or browse the whole strategy section for the operating-model questions that sit around a tool like this.

Want to go deeper? Open Cognibl, read the full proof-of-done guide, see our embedded delivery office, or explore the DATS methodology for placing AI into production.

Written by the Dilr.ai engineering team, practitioners who ship enterprise AI in production. Follow us on LinkedIn for shipping notes, or subscribe via the RSS feed.