Good enough for a ballpark and for the writing around a bid. Not good enough for the number on the tender form — and the reason is not quality, it is that the free tiers lock the automation and the chatbots have no price file.
Key takeaways
There is a version of this question that has a genuinely useful answer, and it is not "yes" or "no". It is: free tools are good at the parts of estimating that involve words and arithmetic you can check, and bad at the parts that involve measuring drawings and knowing prices. Once you see the split, deciding what to use free software for takes about a minute.
Three different things get called free, and they behave nothing alike.
The three kinds of free
A trial. PlanSwift advertises a 14-day trial with full access to every feature and no credit card required. That is a real evaluation, and it is the right way to test a tool on your own drawings — but it ends.
A free tier with the good part gated. Procore's auto-count documentation notes that premium features marked with a star icon are only available to paid Procore customer accounts. You can look at the tool; the automation is behind the paywall. This is the most common shape.
A general chatbot. Free to use, knows a great deal about construction in general, knows nothing about your drawings, your suppliers or your crews.
This is worth understanding mechanically rather than taking on faith. When you upload a drawing to a general-purpose model, the file is handled as pages. OpenAI's documentation on PDF inputs says that for PDFs on vision-capable models, the API extracts both text and page images and sends both to the model. Anthropic's PDF support documentation describes the same shape: each page is converted into an image, the text is extracted alongside it, and the model analyses both.
That is reading, not measuring. Nothing in that process establishes the relationship between a distance on the page and a distance on the building. In a takeoff tool, that relationship is a deliberate step — Procore's drawing scale documentation states that the scale is not configured by default and that setting it is imperative, because it determines the measurement between the plan and the actual project. A chatbot has no equivalent step, so any dimension it gives you is inferred from what is written on the drawing, not measured off it.
The pricing gap is the same story. Canadian construction input costs move quarterly and are tracked in Statistics Canada's building construction price indexes; a general model is not wired to that table or to your supplier's current quote sheet.
Before you paste anything into a free tool, consider what is in it. Tender documents frequently carry confidentiality terms. Drawings identify a client's property. A bid package can contain subtrade pricing that was given to you in confidence.
The Office of the Privacy Commissioner's guidance on AI, privacy and your business is direct about what responsible use looks like: be transparent about how information is used, develop safeguards, and limit the sharing of personal, sensitive or confidential information. The fuller principles for generative AI add explainability and privacy by design. Canada also has a Voluntary Code of Conduct on advanced generative AI systems setting out commitments for the firms building these systems — useful context, but it does not transfer your obligations to them.
None of this makes free tools unusable. It makes the free tier a poor place for a live tender package, and a perfectly reasonable place for a generic question about a construction sequence.
The same federal privacy law that guidance is drawn from is specific about what protecting that information actually requires once it is somewhere. PIPEDA’s Schedule 1 states, as Principle 7 — Safeguards: “Personal information shall be protected by security safeguards appropriate to the sensitivity of the information”, with the methods of protection running from locked filing cabinets to passwords and encryption. A free tier with no enterprise data agreement and no stated retention policy is not obviously providing safeguards appropriate to a client’s address and a subtrade’s confidential price — which is a reason to ask the question before pasting, not after.
None of that changes on a paid tier either; the safeguard obligation attaches to the sensitivity of the information, not to the price of the plan.There is a real list here, and most contractors under-use it.
Writing the words around the number. Inclusions and exclusions lists, scope clarifications, the covering letter, the qualification paragraph explaining what your price assumes. This is drafting, you review every line, and it is where a chatbot is strongest.
Checking arithmetic you show it. Paste your own quantities and rates and ask it to recompute the extension and the totals. You are using it as a second pair of eyes on your own figures, not as a source of figures.
Reading a specification section. Pull out submittal requirements, warranty periods, and what the specification says about a product you had assumed. Then verify the passages against the document.
Unit conversion and sanity checks. Metric drawings with imperial supplier pricing is a routine source of error, and this is exactly the sort of mechanical conversion worth double-checking.
You bid residential renovations and small commercial refits. Roughly a dozen quotes a month, most under $150,000, priced from site visits and photographs rather than full tender sets.
A sensible free-tools setup: the site measurements stay manual, because on a renovation the site visit is the takeoff and no software is going to tell you what is behind the plaster. Your prices live in a spreadsheet you refresh from supplier quotes. A free chatbot drafts the scope-of-work narrative, the exclusions list and the follow-up email. It never sees the client's address or the neighbouring unit's drawings.
When you would pay: when quotes go past about twenty a month, or when you start bidding tendered work with drawing sets, because that is the point where measuring by hand becomes the constraint and a trial of a real takeoff tool will show you the difference in a fortnight.
Not one this article can point to and stand behind. The automated measuring features documented by the major products sit on paid tiers. If you find one that claims otherwise, test it on a job you have already closed out and compare its quantities against what you installed.
For arithmetic, yes. For quantities off a drawing, checking is the same work as doing it, so there is nothing left to save.
Yes, if you spend it on a real job rather than a demo file. Take a project you have already built, run the trial against its drawings, and compare the output to your actual installed quantities.
A 30-minute call is enough to tell you whether AI pays for itself here.
