Continuous AI Competency Training
AI moved fast. Your training never caught up.
terlo senses how your people actually use AI in their day-to-day work, scores it against a live competency model, and builds each person a training pathway matched to their real gaps.
Why terlo
Six ways terlo closes the gap.
AI is moving faster than people's ability to use it well, and nobody can see who's using it carelessly, inconsistently, or not at all. Regulators now expect evidence of workforce AI competency, not a policy document. Generic, one-size-fits-all training doesn't produce that evidence: it bores the confident, misses the at-risk, and burns budget on people who don't need it.
Save training costs
No more paying for courses nobody needed. Budget and time go only where a real gap exists, so the same spend produces measurably more capability — not just a smaller bill.
Build trust in AI
Built as enablement, not surveillance, so people actually engage with it. What someone sees is a pathway built for them — not a dashboard built to watch them.
Target your training
Replace the one-size-fits-all course with a pathway built from what someone actually does. It stretches the fluent and supports the behind, automatically, with no admin overhead.
Evidence workforce skillsets
Move from a policy document to a defensible record. Every score carries a full, explainable evidence trail — the kind of proof regulators and auditors actually expect.
Measured outcomes
Report on competence, not completion. Individual, team, and org-level visibility from the same underlying data, so each stakeholder sees what they're actually accountable for.
Real Usage Visibility
See how AI is actually being used, not how a policy says it should be. Passive sensing across chat assistants, coding tools and copilots gives you real visibility — already proven across Claude, Gemini, ChatGPT and Copilot.
How It Works
terlo executes a value cycle, not a training project.
Sensing, scoring, pathway, delivery, growth — five stages, one continuous loop, running per person across your whole workforce. Nobody has to trigger it, review it, or build a curriculum by hand. It senses real usage, builds what each person actually needs, and re-seeds itself the moment they're done — so the value never stops accumulating.
Frequently Asked Questions
Everything worth asking, answered up front.
Deployment, Privacy & Security
Does terlo store our data?
As a primary option, we deploy a single-tenant instance for each customer into their chosen cloud provider — so your data is 100% owned and controlled by you, in the location of your choice.
What cyber security accreditations can you offer?
The single tenant (customer-owned infrastructure) approach lets us provide bespoke cyber accreditation for each deployment, tailored to what that customer actually needs — ranging from Cyber Essentials Plus, through OWASP ASVS, up to SOC 2 Type II.
Isn't this just surveillance?
terlo is built and positioned as personal enablement, not monitoring. What a person sees is a scorecard and a pathway built for their own benefit — not a dashboard built to watch them. Keeping it feeling that way is something we treat as an active design commitment, not a box we've already ticked.
What data are you actually capturing?
Prompt excerpts for context, not full conversation logs — a deliberate design decision, not a placeholder. That's what makes scoring and pathway generation possible without keeping a full transcript.
Pricing & Commercial
Isn't this just going to cost more, on top of the AI tools we already pay for?
terlo is designed to redirect training spend you already have, not add to it — putting budget toward the people with a real gap instead of a blanket rollout for everyone. In many cases, training savings offset and exceed the terlo implementation and licensing costs.
How do I justify this budget internally, on top of what we already spend on training?
Reframe it: this isn't new spend, it's the existing training budget spent more precisely — the same hours currently going to a generic course get redirected to the people with an actual gap. We don't have a hard £-saved figure yet, since there's no live customer data — lead with the mechanism, not a number.
What's this actually going to save us, in pounds?
Honestly, no measured figure exists yet — this is a demo-phase product. What's real is the mechanism: targeted allocation instead of blanket spend, and the candidate metrics (£/hours reallocated, cost per competency-band uplift) we're tracking for when live data exists.
Product Features
We already run AI training. Why replace it?
Generic courses measure completion, not competence — they can't tell you who's still taking real risks and who's already fluent. terlo isn't "better course content," it's a different category: the training is a downstream output of real usage data, not the product itself.
How is this different from an LMS?
An LMS delivers a course someone picks. terlo builds the course automatically from what that person actually does, and answers "did their competency actually change" — not just "did they complete it."
How do you know you're capturing real usage, not just guessing?
Sensing already works across multiple AI surfaces — chat assistants, coding tools, embedded copilots. A working prototype already senses usage across Claude, Gemini, ChatGPT and Copilot — this isn't a roadmap promise, it's closer to built than most of the rest of the product.
Do we need to install and maintain a new agent?
No. There's already a working prototype for browser-based sensing across Claude, Gemini, ChatGPT and Copilot — this is closer to built than most of the rest of the product, not a future promise.
Training Content
How do you know the pathway is right for someone, not just plausible?
The pathway is driven by that individual's real usage, not a static assessment or self-report. Honest caveat: the competency model's actual skill dimensions and scoring rubric are still being finalised, so "how it decides" is more design intent than finished detail right now.
Will confident, skilled people find this condescending?
The model is designed to differentiate by proficiency — advancing someone already fluent rather than repeating remedial content. That's a design principle the product is built around, not a hope.
Why should someone trust a score they can't see the reasoning behind?
They shouldn't have to. Score explainability is a stated design principle, not an afterthought — a person, or their manager, can see why a score is what it is, not just the number.
How do you know a dip is an AI-skills problem, not something else?
terlo surfaces the usage pattern correlated with an output gap — a strong signal, not a claim of sole causation. It's a diagnostic input a manager still applies judgement to, not an automated verdict.