The Freelance Contract That Almost Owned a Rebrand

The situation

Roughly $165,000 in licence fees, plus a running royalty, sat on the table when Anjali brought the paperwork in for a second look. Her Smiths Falls marketing agency had grown over six years into an established business doing close to $1.8 million a year, mostly building brand packages for small manufacturers and a few larger clients across eastern Ontario. Before any of that, she had spent almost a decade working as a respiratory therapist, and she still called the switch into marketing the riskiest thing she had ever done with a stable career.

The deal on the table came from Sunita, who ran a company that wanted to relaunch its own brand and had settled on licensing an entire content library Anjali's agency had built: logos, a full tagline set, ad copy variations, and a style guide, most of it produced with the help of generative AI tools over the previous two years. Sunita's company would pay the upfront fee, take an exclusive licence to use and adapt the material, and pay Anjali a royalty on any further commercial use tied to future product lines. For Anjali's agency, it was the largest single contract she had signed, roughly equal to a month of ordinary billings.

The library itself had a history Anjali had not thought much about since it was built. Two years earlier, when the agency was smaller and she was still doing most of the production work herself, she had brought in Vikram, a freelance contractor, to set up the AI content workflow: the prompt templates, the brand-voice document, and the subscription structure that everything downstream was generated from. Vikram had since moved on to other work, welding for a local fabrication shop most weeks and taking the occasional freelance contract on the side, and had no ongoing involvement with the agency or any reason to think about the arrangement again.

Anjali's original agreement with Vikram was short, written quickly during a busy stretch, and two years old by the time the licence deal came together. She had never gone back to reread it, and had no particular reason to; the workflow it described had simply become part of how the agency operated. Before signing anything with Sunita's company, she brought the full package to us, mostly as a formality. She expected a quick sign-off so the deal could close on schedule, with the real work already done and only the paperwork left to confirm.

What the review found

The problem was not the AI-generated marketing copy and images themselves. It was the document that had shaped how they were generated in the first place. When we pulled Vikram's original contractor agreement, it turned out he had negotiated, at the time, to keep ownership of the prompt library, the style guide, and the workflow documents he built while setting up the account, on the understanding that he wanted to reuse the same tools for future clients of his own. Anjali had paid for and used everything the workflow produced, but the underlying documents that defined how the AI system generated content were, on paper, still his.

That distinction mattered because the licence deal Anjali was about to sign included a standard warranty: that she owned, or had full rights to license, every asset in the package being handed to Sunita's company. If Vikram's retained ownership of the source documents meant some of the generated content was derivative of material he still controlled, that warranty would not have been accurate. Sunita's company was paying a significant fee specifically for clean, exclusive rights, and a licence built on a warranty that did not hold up would have exposed Anjali to a claim for the fee back, plus damages, the moment anyone looked closely at how the assets had actually been produced.

We also had to work through how much of the finished library actually depended on Vikram's documents versus material Anjali or her staff had built independently since he left. Some assets, like taglines drafted entirely by her team using tools she now controlled directly, were clearly hers, generated with accounts and templates she owned outright. Others, particularly the style guide and several image sets generated from prompt structures Vikram had written, traced back to work product his contract still covered. Roughly a third of the library fell into that second category, which was more than Anjali had expected once we laid it out asset by asset.

None of this had surfaced earlier because nobody had connected the two documents. The contractor agreement lived in an old email folder from years before; the licence deal was being negotiated fresh, with its own team and its own timeline. It took a direct request to pull both files and read them side by side before the gap became visible, and it became visible only three weeks before the scheduled closing date, with a fee already agreed and a launch date already announced to Sunita's own customers.

What we did

1. Mapped the content library against its sources. We went through every asset in the package Anjali intended to license, including logos, copy, images, and the style guide, and sorted each one by which underlying document or workflow it had come from, so we knew exactly which pieces carried ownership risk and which did not before touching the licence agreement itself.
2. Reread Vikram's original contractor agreement line by line. The document was short and informally drafted, and its ownership language was easy to miss on a first pass. We confirmed it gave Vikram ongoing rights to the prompt library and workflow documents specifically, not to any content Anjali's team generated using tools they controlled independently afterward, which narrowed the exposure considerably once it was clear.
3. Paused the licence agreement before it was signed. Rather than let Anjali warrant ownership she could not fully back, we advised holding the signing until the gap was resolved, and drafted language for her to use with Sunita's company explaining a short, specific delay without disclosing more detail than necessary to keep the deal on track. A warranty that later proved false would have cost far more, in money and in trust, than a well-managed pause did.
4. Contacted Vikram to negotiate an assignment. We reached out on Anjali's behalf to buy out his retained rights in the specific documents the licence deal depended on, rather than trying to argue the point or route around it, since a clean assignment was faster and far less risky for both sides than a dispute over informally drafted terms that neither of them had really turned their mind to at the time it was signed.
5. Negotiated and drafted the assignment agreement. Vikram agreed to assign his rights in the prompt library and style guide documents for a payment in the low five figures, reflecting the narrow scope of what he was actually giving up rather than the value of the whole deal downstream, which kept the negotiation quick and cooperative rather than adversarial, since Vikram had no real incentive to hold out once he understood exactly what was being asked of him.
6. Rewrote the warranty and representation clauses in the licence agreement. Once the assignment was in hand, we tightened the ownership warranties in Anjali's agreement with Sunita's company so they matched exactly what she now held clear title to, closing the gap the review had found and giving both sides language they could actually rely on if the material was ever challenged after closing.
7. Advised on the agency's future contractor agreements. To prevent the same issue recurring, we recommended standard ownership language for any future freelance or contractor work involving Anjali's AI-assisted content pipeline, so any tools, templates, or documents built for the agency belonged to the agency from the moment they were created, regardless of who did the building, closing the exact gap that had put the licence deal at risk.
8. Briefed Anjali on how to describe the delay to Sunita's company. A short, unexplained pause this close to closing risked reading as cold feet or a pricing renegotiation. We gave Anjali plain language she could use that was accurate without disclosing the ownership gap itself, which kept Sunita's trust in the deal intact while the assignment with Vikram was finalized behind the scenes.

The outcome

The licence deal closed roughly three weeks later than originally planned, once the assignment with Vikram was signed and the warranty language was corrected to match it. Sunita's company received the content library it had agreed to pay for, with ownership behind it that actually matched what the agreement promised, and the deal proceeded on the commercial terms both sides had negotiated from the start, without any reduction in fee or royalty.

The delay and the assignment payment together cost Anjali roughly $14,000 beyond her legal fees, the sum of Vikram's buyout and some minor renegotiation of the closing timeline with Sunita's company to account for the short delay. Set against a $165,000 licence fee and an ongoing royalty on future use, it was a manageable cost, and nothing about the underlying deal changed for either side once the paperwork caught up with the facts.

What made the outcome a prevention rather than a repair was timing. Had the gap surfaced after closing, once Sunita's company had already begun using the material commercially in its own rebrand, Anjali would have been negotiating from a position of exposure rather than choice, likely at a worse price and with a warranty claim already in motion against her. Because the review happened before signing, the fix was a straightforward commercial negotiation with someone who had no reason to be difficult about it, rather than a dispute conducted under pressure after the fact.

Anjali also came away with a cleaner contractor template for future work, which meant the same gap could not reopen the next time she brought in outside help to build or expand the agency's AI-assisted production tools. She now runs any new licensing or sale agreement past that template before quoting a fee, rather than after a client has already agreed to one.

What you can learn from this