.png)
-42%
Average token reduction
per prompt
€ 180
Avg monthly AI cost saved
per 10-person team
2.4 KG
CO₂ avoided per team
per month
About us
Insusty's Poca the Prompt Partner sits inside your team's workflow, refines every AI prompt in real time, cuts unnecessary token cost, and tracks the environmental savings automatically.
The Problem
Your team is using AI.
Nobody is optimising it.
Every unoptimised prompt wastes tokens, inflates your AI bill, and adds to your company's carbon footprint. IT teams have no visibility. Finance has no control. CSR has no data.
Companies overspend on AI tokens by 30–50% due to verbose, unstructured prompts that could be refined in seconds.
AI usage is a growing part of corporate Scope 2 emissions. Most companies can't measure it, let alone report it.
IT and CSR teams have no dashboard, no data, and no proof of responsible AI use for CSRD reporting.
of enterprise AI prompts contain redundant tokens that inflate cost and energy use
AI energy consumption is projected to triple in European companies by 2027
Product roadmap
Built to be the most profitable AI sustainability platform in Europe.
Poca is the first product. The platform expands into a full suite of AI sustainability tools for enterprise.
Now · 2026
Poca Web App
- Prompt refiner + 3 metrics
- User dashboard & history
- Per-seat B2B pricing
- IT admin reporting
Q3 2026
Chrome Extension
- Poca floats on every page
- Works inside ChatGPT, Claude, Copilot
- Enterprise deployment via IT policy
- Team usage analytics
Q4 2026
Insusty Platform
- Poca weekly email nudges
- Employee check-in dashboard
- Auto PDF proof report (CSRD-ready)
- Scope 2/3 AI emissions tracking
2027
AI Sustainability Suite
- Multi-model cost optimisation
- AI procurement advisory
- ESG reporting integrations
- Enterprise API & white-label
FAQ
Works with ChatGPT, Claude, Gemini, Copilot and more.
Poca works alongside any AI tool. As a web app you paste your prompt in directly. The Chrome extension (coming Q3 2026) will float inside ChatGPT, Claude, Gemini, and Microsoft Copilot : no workflow change needed.
Estimated using Ecologit's methodology for AI energy consumption.
We use Ecologit to estimate the carbon footprint of AI prompt usage. Ecologit measures the energy consumption of AI model inference and applies the carbon intensity of the electricity grid to calculate CO₂ equivalent emissions. The figure is clearly labelled as an estimate and based on Ecologit's published methodology. We share this openly so your CSR team can validate it for CSRD reporting.
Web app: zero setup. Extension: central IT deployment via policy.
The web app requires zero IT setup : share a link and your team is using it in minutes. The Chrome extension can be deployed to all company browsers centrally via Google Workspace or Microsoft Intune, with no action required from individual employees.
Team usage, tokens saved, cost savings, CO₂ avoided, per-user breakdown.
Team-wide prompt volume, total tokens saved, cumulative cost savings, CO₂ avoided, and per-user breakdowns. The monthly PDF report is auto-generated and ready to attach to your CSRD or ESG documentation.
No , any team of 5+ using AI will see immediate ROI.
Any team using AI tools regularly "even 5 people" will see immediate cost savings that exceed the subscription price. CSRD reporting features become more relevant at 250+ employees, but the prompt optimisation ROI is instant at any size.
No , Poca makes your policy measurable and provable.
Your policy sets the rules. Poca makes following them measurable and gives you the data to prove compliance internally and externally. Think of it as the reporting layer on top of your existing governance.
Free 2-week pilot. No credit card, no commitment.
We set up a free 2-week pilot for your team with full access. No credit card, no commitment. You see the savings first, then decide.
Interface in English and French. Prompt refinement works in any language.
The interface is currently in English and French. Poca understands and refines prompts in French, English, Spanish, German, and more , it works in whatever language your team writes in.
ChatGPT is built to answer you. Poca is built to save you money and prove it.
Asking an LLM to shorten your prompt is like asking a petrol station to help you drive less. It was not designed for that goal and it has no incentive to do it well. Poca is different in four ways. First, it is optimised specifically for token efficiency and sustainability, not just brevity, the refinement logic is built around cost and carbon reduction. Second, it teaches you to prompt better over time by showing the reasoning behind every change, so your team gradually needs less correction. Third, it accumulates every saving into a reportable dashboard your IT team sees the total, not just one prompt at a time. Fourth, it gives CSR and finance teams company-level proof they can attach to CSRD reports and board presentations, something no chat window will ever produce.
Stored only with explicit consent, and it benefits your team directly.
By default we store only metadata : token count, cost saved, CO₂ estimate, never prompt content. With company admin consent, anonymised prompt patterns are stored to build your organisation's AI usage intelligence: what your team uses AI for, where inefficiencies cluster, which departments prompt well and which need coaching. This data feeds back into your IT dashboard and helps Poca improve its refinements over time. All storage is GDPR compliant, EU-based, governed by a Data Processing Agreement, and can be disabled or deleted at any time.
