The default nobody checked
Most firms picked their slide tool by accident
Most consultancies are running whatever presentation tool someone on the team stumbled across six months ago. A LinkedIn post, a quick trial, and suddenly that tool is the default. Nobody checks whether something better exists. Nobody has time to run a proper test.
The market has become genuinely hard to navigate. Canva has one. Gamma has been around a while. ChatGPT launched one recently. Microsoft Copilot is built into PowerPoint. Google has its own inside NotebookLM. And newer entrants — Kimi, Perceptis, Manus — keep arriving.
If you picked one six months ago and stuck with it, you have no idea whether you made the right call. The real problem is not the number of options — it is that most tools are built for demos, not consulting delivery. They look impressive in a post and fall apart when you hand them a real brief. So I tested the ten most popular tools properly.
Three real briefs
The decks that actually matter in consulting
I designed the tests around the decks that matter in real work, not generic demos.
- Client proposal. A mid-size firm pitching a six-figure operational efficiency programme: phased delivery, projected savings, eight specific slide requirements. High complexity, numbers that matter — the brief a senior consultant hands a junior before a pitch.
- Quarterly business review. Over twenty data points — revenue versus target, margins, utilisation, pipeline, win rate, NPS. The kind of deck that falls apart fast if a tool cannot handle data-dense content coherently.
- A one-line brief. “Create a five-slide presentation about why UK consulting firms should adopt AI agents in 2026. Include risks and benefits. Keep it practical, not hype.” This existed to see what tools do when you give them almost nothing. A few surprised me. Most did not.
Thirty presentations in total — one from each tool, per test case.
Scoring the output
A hybrid model: AI on what it can judge, me on the rest
Using AI alone to evaluate everything was tempting, but AI does not know what a client-ready deck looks like — it cannot read the finish or apply the gut check that comes from sitting in a client meeting. Reviewing all thirty myself was the other option, but consultants bill for their time. So I built a hybrid.
The AI scored three dimensions:
- Storyline and structure. Does the deck tell a coherent story, or is it just slides in a pile?
- Action titles. Does each title state an insight — “Revenue missed by £300K, win rate is the root cause” — or just a label like “Financial Performance”? If you cannot follow the argument from the titles alone, the deck is not working.
- Content specificity. Did it use the actual names, numbers, and details from the brief, or is it generic enough to swap the client name and change nothing?
I scored two dimensions:
- Data and visualisation. Are the charts the right type, supporting the narrative rather than decorating it?
- Client-readiness. The gut check. Could I send this, and how much work sits between the output and something I would present?
Editability was scored separately — and it is make or break. A deck you cannot edit is knowledge trapped in an image file.
Before AI can do useful work in your firm, it needs to produce outputs your team can actually touch, change, and build on — and that is where most tools fail.
For the AI pass I built a detailed scoring prompt for Claude Sonnet 4.6 and gave it access to all thirty decks via Claude Cowork, which works directly with your local file system. I hit usage limits twice while refining it — which tells you something about the volume involved. The AI evaluation carries a weighting of six out of ten; my human scores add the remaining four.
The leaderboard
What ranked where, and why
The rankings below run worst to best.
- #10 Pitch.com. The lowest scorer overall. Weak content across all three cases, and PPTX export is paywalled on the free tier. Without editability it scores essentially nothing useful. Not worth your time.
- #9 Manus 1.6 Lite. One out of three for content, one for editability — a total AI score of two out of six. Not a credible deck-building tool right now.
- #8 NotebookLM. A special case. The client proposal scored a perfect six out of six; the QBR and one-liner scored four each — the second-highest content score of any tool. But editability is zero: every slide exports as a PNG inside the PowerPoint. You cannot change a word, a number, or the brand. Were the slides editable, it would be near the top.
- #7 Canva. Content 1.3, editability two — total 3.3. A popular design tool, but not built for consulting-grade slide work.
- #6 Gamma. The name most associate with AI presentations. Content 1.5, editability two — total 3.5. Output reads like a branded infographic. In fairness this used default settings and a single prompt; with brand templates it may differ. But it is not the default answer it is assumed to be.
- #5 Microsoft Copilot for PowerPoint. Content 1.5, editability three out of three — native PowerPoint gives the best file compatibility tested. Total 4.5. Editability is strong, but content quality lags what the platform should manage.
- #4 Kimi. A Chinese tool most UK professional-services people will not have heard of — and it beat both Gamma and Copilot. Content two, editability three — total five out of six.
- #3 and #2 ChatGPT and Perceptis (tied). Both scored 5.2 out of six. Perceptis was unknown before this test, so matching an industry-recognised tool earns it second.
- #1 Claude Design. Content 2.7 out of three — the highest of any tool — and editability three out of three, for a total 5.7. It also has a PowerPoint add-in giving full editability inside your existing workflow. For decks that need to be accurate, client-ready, and editable, this is the one that performed.
Adding my own human scoring for visual design and client-readiness — worth four further points — did not change any tool’s position.
What to take from it
The tool removes the drag, not the consultant
None of these tools replace a consultant who knows what a good deck looks like. The best ones save you hours on a first draft. The worst ones cost you hours cleaning up after them.
The right tool removes the drag between the brief and the first draft — which is where senior time goes. That is what making AI work for your team actually looks like in practice. Now you know which is which.