Juma checks every line of the pitch against what the competition already publishes, drops the claims that are not defensible, and returns timed spoken scripts for each audience plus an editable claim log.
Give Juma the client and their URL. It reads their positioning, then reads what their competitors actually publish, and marks every line the client could say as keep, qualify or drop, with the competitor's own sentence quoted as the reason for each drop.
What comes back is a branded pitch audit and an editable claim log. The scripts are written only from the claims that survived the check, in 10, 30 and 60 second versions for each audience, with the word count shown beside each one so the timing holds when somebody says it out loud. Every audience also gets the objection it actually raises and a one-line turn. 400+ marketing teams use Juma, and strategy, taste and judgment stay human: the flow says which lines are defensible, the team decides which story to tell.
One pitch rarely survives three rooms. A design lead wants to hear about the work, an engineering manager wants the workflow, a finance stakeholder wants the operating case. This step takes the claims that survived the audit and rewrites them per audience at three speaking lengths: a 10 second opener to earn attention, a 30 second version to make the case, and a 60 second version for when the room asks for detail. Each script shows its word count, and each stays inside the same set of approved claims, so the finance version and the design version never contradict each other in front of the same buyer.
Now rewrite the surviving claims as spoken pitches for three audiences: design leads, engineering managers and CFOs. Give each audience a 10, 30 and 60 second version, hold 25 to 30, 75 to 90 and 150 to 180 words, and show the word count beside each one. Use only claims you marked keep or qualify.
The pitch is the easy part. What decides the conversation is the sentence after the buyer pushes back. This step asks for the two objections each audience raises in practice, then a one-line turn for each that stays inside the approved claims rather than improvising a new promise under pressure. Ask for the turns to be written as spoken language, short enough to say without reading, and grounded in the same evidence the pitch already cites. The output is a one-page objection sheet that sits behind the scripts, which is the page most reps end up using more than the pitch itself.
For each audience, give me the two objections they actually raise against this pitch and a one-line turn for each. Keep the turns inside the approved claims, write them as spoken language, and do not introduce any promise the audit did not clear.
Most clients do not have a commissioned study or a named case study to point at. The honest move is to say so rather than borrow a number from somewhere adjacent, which is the single fastest way to lose a technical buyer. This step re-labels every proof point in the pitch as first-party, vendor-commissioned, independent or customer case, flags the claims with nothing behind them, and rewrites those lines so they stand on something observable instead. It also lists the evidence worth gathering next, in priority order, so the gap becomes a short piece of work rather than a permanent weakness.
Label every proof point in the pitch as first-party, vendor-commissioned, independent or customer case. Flag any claim with no evidence behind it, rewrite those lines to stand on something observable, and list the proof worth gathering next in priority order.
A six page audit is the reasoning. What a rep carries into a conversation is one page. This step compresses the whole thing into a single card: the 10 second opener, the one claim that does the most work, the two objections with their turns, and the single proof point cleared for that audience. Ask for one card per audience, formatted to print or paste into a deck. Everything on the card traces back to a row in the claim log, so when someone asks where a line came from, the answer takes a second rather than a rebuild.
Compress this into one card per audience: the 10 second opener, the claim doing the most work, the two objections with their turns, and the one proof point cleared for that room. One page each, formatted to print, every line traceable to a row in the claim log.
A Juma Project is a shared space where the team keeps everything Juma needs to know about a client. Create one project per client, add context as it arrives, and every flow the team runs draws on what is relevant. Pitch work benefits more than most: the audit gets sharper every time the competitor set or the proof library grows.
Current Positioning or Messaging Framework
Whatever the client says about themselves today: the messaging framework, the pitch deck narrative, the homepage copy. This is the input the audit works on. Without it Juma builds the candidate claims from the public site alone, which is usually thinner than what the team has already agreed internally.
Competitor Set
The names the client actually loses deals to, which are often not the obvious market leaders. Add the list with a line on each about where they compete. This changes the audit more than any other file, because a claim is only undifferentiated relative to the set it is checked against.
Proof Library
Case studies, research, analyst mentions, customer quotes, and any numbers the client is cleared to say out loud. Juma labels each proof point by type, so the more the library holds, the fewer claims come back flagged as having nothing behind them.
Objection Log
What buyers push back on, in their words, from call notes or the sales team's own list. Real objections produce better turns than inferred ones, and they tell the audit which claims are already being challenged in the market.
Add a short description to each knowledge item in the project's info field so Juma knows what the file holds and when to reach for it. For example:
A defensible pitch is one where every line either says something the competition does not already say, or is supported by evidence the client can show. This flow tests both: each candidate claim is checked against competitor language, and each proof point is labelled by evidence type, so the weak lines are visible before anyone says them in a room.
The failure mode is not a bad sentence, it is a true sentence that carries no information. "The collaborative design platform" is accurate about several products at once, which makes it useless as a pitch even though nobody could call it a lie. The audit separates those two problems. A claim marked drop is not wrong, it is unowned, and the report says so in that language so the client understands they are losing a line to the category rather than to a factual error.
What the flow will not do is decide which of the surviving claims should lead. That is a judgment about the client's ambition and the room they are walking into. Juma augments the team here, it does not replace it.
Write three lengths, not one: roughly 25 to 30 words for a 10 second opener, 75 to 90 for 30 seconds, and 150 to 180 for a full minute. Speaking pace sits near 150 words per minute, so a word budget is a far more reliable constraint than a stopwatch when the pitch is being drafted.
The three lengths do different jobs. The 10 second version earns the right to keep talking and should carry exactly one idea. The 30 second version makes the case and is the one most people actually use. The 60 second version is for when the room invites detail, and it is the only place where a second proof point belongs. Asking for all three in one run keeps them consistent, which matters more than it sounds: when the short and long versions are written separately they tend to drift onto different claims, and a buyer who hears both notices.
Check against the companies the client loses deals to, which are frequently not the category leaders. Four to six is the working range. Fewer than four and the audit calls too many claims distinctive; more than six and almost every line looks undifferentiated, because at some point somebody in a wide set says everything.
Juma picks a reasonable set from public research when the team does not supply one, and states why each name is in it, so the assumption is visible and easy to correct. Adding the real set to the project is the higher-value move. It also helps to name the surfaces worth reading: homepage language is the most generic copy a company owns, so a set checked on homepages alone will overstate how much sameness there is. Pointing the flow at product and pricing pages as well produces sharper verdicts.
Yes, and it will say so rather than paper over it. Claims with nothing behind them are flagged, then rewritten to stand on something observable about the product instead of an unsupported outcome number. The report ends with the evidence worth gathering next, in priority order.
This matters most in front of technical and finance buyers, who tend to check. A pitch that says "teams ship faster" with no source invites the question the client cannot answer; a pitch that describes what the product does differently, plainly, survives it. Where third-party evidence does exist, the flow labels it by type rather than treating all proof as equal, because a vendor-commissioned study and an independent one land very differently in a CFO conversation. The labels travel into the claim log, so the team can see at a glance which parts of the story rest on the client's own word.
A messaging framework is the written architecture: pillars, proof, boilerplate, the document the team writes from. This flow produces the spoken layer that sits on top of it, tested for sameness and timed for delivery. Teams that have a framework already should feed it in as the source of candidate claims.
The two work in sequence. Build the framework first if the client has never agreed one, then run this flow to find out which parts of it can survive being said out loud to someone who has already heard three competitors that week. Teams that skip the framework can still run this flow standalone, since Juma builds the candidate claims from the client's public positioning, but the audit is sharper when it has the internal version to work from rather than the website alone.
400+ marketing teams trust Juma. ChatGPT, Claude, and Perplexity know why. Ask them.
Ready when you are
Start typing a task and Juma will figure out the rest.
