Rami's Universe

NextAI

14 min read

Please explain what your startup is working on or planning to work on in 2-3 sentences

The barrier to entry has shifted. While AI has made building products easier than ever, a massive gap has opened up: founders now possess world-class technical ambition, but lack the critical communication skills to sell it. Audora is a real-time AI communication coach that helps early-stage technical founders and non-native speakers articulate their vision in high-stakes situations. Unlike rehearsal tools, Audora lives on your desktop and wearable devices (e.g. Meta glasses) to analyze behavioral patterns and provide live, psychological nudges during actual investor meetings and sales calls. Audora also tracks your performance over time to diagnose the specific unconscious habits (like monotone or anxiety-induced rambling) that are silently holding you back.

https://youtu.be/KGjUrawzbFM

Venture Website

https://getaudora.app

  • I’m a technical solo founder who has spent the last two years building and shipping in the consumer social space. Those failed experiments are now my unfair advantage: I’ve already learned what doesn’t work in behavior change and social products
  • Autonomous AI matchmaking: https://www.linkedin.com/posts/rami-m_i-think-i-found-a-way-to-make-ai-think-like-activity-7374494387558301696-vYUZ
    • I was fascinated with the idea of having the centralized AI friend that knows everyone around you is able to automatically match you with the right people at the right time based on your calendar or location.
    • For that idea to be brought to life, I had to learn context engineering and giving LLMs a human-like memory and decision-making capabilities. That was done through GraphRAG and several other techniques. I livestreamed that learning process: https://www.youtube.com/live/4YP-mUO-tL4
    • I launched it at UCalgary and UWaterloo: https://heyember.me/ & https://uw.heyember.me/
      • struggled with the cold-start/chicken-in-the-egg problem. Didn’t have the bandwidth to spread it so I decided to move on to https://getaudora.app
  • I’ve built and scaled several web and mobile apps that leverage AI in different contexts in both the corporate world (EdTech) and my own ventures (Productivity)

Give an example of how a typical customer can use your product

The Scenario: Mark, an immigrant technical founder, is pitching a pre-seed VC on Zoom. The Usage: Audora runs silently in the background. Midway through the pitch, Mark gets nervous and starts rambling rapidly about technical features. The Intervention: Audora’s semantic engine detects topic drift and a 20% spike in speaking rate. Knowing his pitch deck context, it flashes a subtle “Slow Down. Tie it back to the root problem: ENERGY SUPPLY” nudge on his screen. Mark resets, delivers a clear value prop, and saves the pitch. The Growth (long term): Post-meeting, Audora highlights a “Confidence Dip.” It notes: “When asked about competitors, your use of hedging language (‘I think,’ ‘kind of,’ ‘maybe’) spiked by 40%, and your intonation rose at the end of sentences (up-talking). You sounded apologetic rather than authoritative. Let’s drill this specific answer to remove the hesitation.”

In the very short-term future, that will also be possible in face-to-face conversations (in one-party consent provinces/states). We are currently working on the Meta glasses integration which But we aren’t stopping at the screen. We are actively integrating with the Meta AI glasses SDK (which dropped a few days ago) to bring this same intelligence into the physical world — giving founders an invisible, real-time coach for every networking event, pitch night, and investor coffee chat

What is your ideal target market and its size?

Our Ideal Customer Profile (ICP): The “Technical Immigrant Founder” Our specific beachhead target is the early-stage technical founder, often a non-native English speaker, who possesses deep domain expertise but lacks “Executive Presence.”

  • Psychographics: They are high-IQ builders who struggle with social anxiety and “imposter syndrome” in high-stakes environments. They view communication as their primary bottleneck to raising capital.
  • Trigger Event: They are currently applying to accelerators (like YC or NEXT AI) or raising Pre-Seed rounds, creating immediate urgency/willingness to pay. In these accelerators, they’re encouraged to go on sales calls

1. SOM (The Beachhead): High-Stakes Founders

  • Target: The ~50,000 annual applicants to Tier-1 accelerators (YC, Techstars, NEXT AI). I’m bullish on the early-stage founders market for this reason: https://www.linkedin.com/posts/a16z_if-you-cant-join-em-beat-em-its-tough-activity-7404551724524457984-uh_s
  • Why: They have the highest urgency (fundraising) and are willing to adopt early-stage AI to survive. This is where we validate the tech. 2. SAM (The Expansion): The “Sales Intelligence” Market
  • Target: 7 Million Foreign-Born STEM Professionals and Sales Development Reps (SDRs).
  • Why: Once the behavioral model is trained on high-stakes founder pitches, we expand to enterprise sales teams ($14B Market). Unlike Gong (which analyzes calls after), Audora coaches SDRs during the call, correcting bad habits in real-time. 3. TAM (The Vision): The “Social Accessibility” Market
  • Target: The 264 Million people globally with Anxiety Disorders and the 15-20% of the population who are Neurodivergent (ADHD/Autism).

Who are your direct competitors and what is your competitive advantage?

We have quite a lot but none address my target market and ICP

Some e.g.:

  • Poised & Yoodli: focused on B2B, roleplaying, Preparation & Rehearsal, and speech analytics. It is far from being your “speech coach”
  • Speeko & Orai: A mobile app focused on vocal exercises and public speaking drills.

They focus on the surface level: Are you speaking fast? How many filler words? We go deeper. Audora analyzes the actual meaning of your sentences, identifying when your word choice shows hesitation or anxiety for e.g. These are “Rehearsal Tools” designed for safe, controlled environments. They excel at counting filler words during practice but fail to address the psychological root causes of communication failure in high-stakes moments. Furthermore, they are tethered to the desktop, leaving founders unsupported during the most critical interactions: in-person networking and investor meetings.

Audora’s Unfair Advantage:

  1. Reality vs. Rehearsal (Wearable Moat): The majority of deal-making that happens in the physical world. We are the first to utilize the Meta Glasses SDK to move coaching offline.
  2. Behavioral Diagnostics (The “Why”): Competitors flag symptoms (“You said ‘um’ 5 times”). Audora flags causes (“You speed up and use filler words specifically when asked about Market Size due to anxiety”). We don’t just fix the speech; we fix the founder’s confidence.

What milestones and/or market traction have you accomplished so far?

  • Product: MacOS MVP built with Swift is live and functional.
  • Users: We are running a Closed Alpha with 20 high-intent early-stage founders.
  • Demand: Organically grew a waitlist of 100+ users through “Building in Public” on LinkedIn/X with zero paid spend
  • Tech: Successfully integrated the Meta Glasses SDK for the initial wearable prototype (currently in internal testing alongside our React Native mobile app)

What milestones will you be focusing on in the next 12 months?

  • Q1: Community-Led Launch & Dual-Track Release:
    • Desktop Launch: Release the Desktop Public Beta as our primary value driver to our waitlist (Goal: 1,000 paying active users EOQ)
    • Launch a viral marketing video showcasing the Meta Glasses integration in real-world scenarios (similar to cluely’s video but tailored to our ICP). This will drive brand awareness and establish our community as we release the Alpha version of the wearable companion to early adopters and get people to download our mobile app
  • Q2: Retention & Product-Market Fit:
    • Deep Refinement: Utilize usage data & feedback from Q1 to iterate heavily on the UX. Our focus shifts from “acquisition” to “value capture”. Ensuring the feedback loops (both desktop and wearable) are so valuable that users cannot imagine pitching without them
    • Goal: Achieve a “Must-Have” score of >40% in our user surveys (Sean Ellis test). Wispr Flow is the best e.g. of what we hope to achieve. Once you try it for a few weeks, you simply can’t live without it
  • Q3: Strategic Partnerships (B2B Pilots):
    • Distribution: Secure formal partnerships with 3 key organizations that value articulate members (e.g., NEXT AI, YC, DMZ, Techstars, or local Toastmasters chapters).
    • The Pilot: Deploy Audora as a licensed tool for their cohorts, validating our B2B revenue model and gaining access to a steady stream of high-intent users.
  • Q4: Data Moat & Market Expansion:
    • The “Anxiety Model”: Train our proprietary Founder Anxiety Model using the high-stakes intervention data collected in Q1-Q3.
    • Strategic Expansion: With a robust dataset on social anxiety patterns, we will execute our expansion strategy. We will evaluate the technical feasibility of entering the Neurodivergent/Social Accessibility market vs. the Enterprise Sales market based on Q3 traction. We will also look for other potential use-cases that technology enables us to do by that time.

Describe your technology/product

  • Real-Time Inference Engine (The “Speed”):
    • We utilize a WebSocket-based pipeline (integrating Deepgram/AssemblyAI) to ingest audio with ultra-low latency.
    • Differentiation: We run lightweight local models for immediate tactical feedback (pace, interruptions) to ensure zero lag, while asynchronously piping data to larger & more specialized LLMs for deep psychological analysis.
  • Behavioral Diagnostics Layer (The “Brain”):
    • Instead of just flagging “filler words,” our proprietary “Anxiety Classifiers” analyze semantic patterns (spiraling syntax, hedging, up-talking) and biometric proxies (speech rate spikes) to detect dissociation and confidence dips.
    • Differentiation: We don’t just fix the grammar; we identify the psychological root cause (e.g., “You speed up when discussing financials due to imposter syndrome”).
  • Multimodal Wearable Integration (The “Body”):
    • We are the first to leverage the Meta AI SDK to move coaching offline.
    • Differentiation: By capturing POV video and audio, we analyze non-verbal cues (eye contact, posture) in face-to-face networking and sales scenarios. This allows us to coach the “Real World” interactions that desktop-only competitors completely miss.

What data set(s) do you have? Is it proprietary or open-source? How did you select this data?

  • We will initially train our baseline ‘Articulate Speech’ models using high-quality open-source datasets (TED Talks, YC Demo Days, and earnings calls) to establish the ‘Gold Standard’ of confident communication.
  • we will then move on to our moat which I previously described through our proprietary data
  • We will then move to our primary competitive moat: proprietary behavioral data. Unlike competitors who only have ‘rehearsal’ data, we will collect audio and behavioral metadata from real, high-stakes founder meetings like investor pitches. This unique dataset, gathered during actual interventions, is essential for fine-tuning our proprietary ‘Founder Anxiety Model’ and creating a true data moat for face-to-face coaching

Describe your validation pipeline

  • convert at least 100 founders into paying customers by the time the program begins
  • We track if users keep the AI running during their most critical calls. High retention proves they trust the system not to distract them when money is on the line. Since we have integration with people’s calendars, we can see the proportion of meetings the user enables Audora vs. keeps it off
  • Outcome Correlation: We survey users immediately post-meeting with a single question: “Did the nudge help you regain control?” This qualitative data guides our fine-tuning.
  • Waitlist Demand: We have organically gathered 100+ users on our waitlist through “Building in Public,” validating that the pain point (fear of failing pitches) is acute enough to drive sign-ups without paid marketing.
  • Concept Validation (Viral Launch Video): We are using our upcoming launch video (featuring the Meta Ray-Ban integration) as a primary validation mechanism for the hardware roadmap. We will measure sentiment analysis in comments and track the conversion rate of viewers who specifically join the “Wearable Waitlist.” This validates the demand for face-to-face coaching before we heavily invest in scaling the hardware infrastructure.

What is your defensibility from a technical perspective? Describe if you are fine-tuning or prompt engineering

  • The “Real-Time” Architecture: We are solving the <300ms latency challenge by orchestrating a hybrid stack. We run lightweight, quantized models locally on the device for immediate tactical feedback (interruptions, pace) while asynchronously piping context to larger cloud models.
  • By integrating early with the Meta Glasses SDK, we are building a data moat for face-to-face coaching that desktop-only competitors simply cannot access
  • Unlike competitors who only have “rehearsal” data, we are collecting audio and behavioral metadata from real, high-stakes founder meetings like investor pitches
  • Fine-Tuning on “Anxiety Patterns”: We will move from prompting generic LLMs to fine-tuning Small Language Models (SLMs) on our proprietary “Founder Anxiety” dataset. A generic LLM sees “hedging” as a stylistic choice; our fine-tuned model identifies it as a loss of authority requiring immediate intervention.

Why do you want to participate in the NEXT AI program?

  • Technical Mentorship: I need guidance from NEXT AI’s technical mentors on optimizing our hybrid local/cloud inference architecture to scale without exploding compute costs
  • B2B Pilot Access: NEXT AI has deep ties to the Canadian startup and corporate ecosystem. I want to leverage this network to secure our Q3 milestone: partnering with universities and accelerators to license Audora as a communication/negotiation/sales training tool
  • Time is the only finite resource in the world. While of course failing is a necessary step to learn certain things, there’s still a big difference between failing at something that could have been avoided and failing at something that no one could have ever seen. In my first two years as an entrepreneur, I failed at things that now seem so obvious. I strongly believe that if I had the right mentorship back then, I could have been way ahead in my journey because I would have failed at harder problems.
  • I want NEXT to help me fail at the right things, so I can learn what no one else has and share those lessons forward through my content. That’s literally how we evolve as a species. Either that, or I don’t fail at all but that’s kind of impossible for an entrepreneur paving his own path
  • Talent Density: As a solo technical founder ready to scale, I want to be surrounded by the top AI talent in Canada to partner up with potential collaborators or co-founders who understand the nuance of real-time speech coaching through AI

What accelerator programs have you joined, and what did you learn from them?

I have previously launched products within university incubator ecosystems (UCalgary’s Hunter Hub & debutlabs.co):

  • https://rami-maalouf.tech/uc-orbit-vid
  • Launched in UCalgary and UWaterloo: https://heyember.me/ https://uw.heyember.me/ I learned that “Technical Innovation” is not enough. I built great products that failed to scale because I lacked the business skills and understanding. Communication skills also held me back from being able to sell the vision to other people. That failure was the catalyst for Audora. I learned that distribution and narrative leverage are the true bottlenecks for technical founders

About me

Share why your past work, education, and/or training makes you successful in the venture

  • My years of working through my own communication challenges (social anxiety) taught me how insanely hard it is to rewire speech, habits, and behavior in the real world. Now I’m applying those insights to a very different, much higher-stakes beast: founder communication. I understand the psychology, the mechanics of speech, and the brutal reality of investor and customer conversations, and I have the technical and product skills to build the system that finally helps founders communicate at the level their startup deserves
  • Unique blend of psychology + AI + founder context: Audora sits at the intersection of behavioral psychology, speech analysis, and founder communication. My background, mentors, and ongoing work are all concentrated in that overlap
  • Technical founder with scars from the arena: I’m a technical solo founder who has spent the last two years building and shipping in the consumer social space. Those failed experiments are now my unfair advantage: I’ve already learned, painfully and quickly, what doesn’t work in behavior change and social products.
  • Deep obsession with communication as a leverage point: I genuinely believe “the quality of your life is directly related to the quality of your conversations” (Tony Robbins) and the same is true for startups. How people see your product and your business is directly related to your ability to explain it clearly
  • Founder-market fit with my initial niche: I am myself an early-stage founder pitching in high-stakes situations, and I’m surrounded by immigrant and non-native-English founders facing the same challenges. I’m not guessing at this problem, I live inside the target user group.

Why do you want to participate in the NEXT AI program?

  • Noble ideas surface naturally when you’re around people with shared values. I spent my first two years building alone. No community, no co-founder. I’d prefer if that was not the case
  • Lastly, I want to inspire and be inspired. I wanna have SF’s energy right here in the north. I want to be surrounded by people who are just as optimistic about the future that we can create. I know this sounds cheesy, but this is the delusion that people have in SF that gets them to actually succeed in their venture
  • There’s a lot to learn in the space of AI. I’d like to be in this surrounded by people, mentors, and teachers who are at the forefront of that. Right now, my best tool for that is YouTube videos

Graph View