By NS Ramnath & Charles Assisi

Many times in the last few years, we started noticing strangers in our Founding Fuel Masterclass Zoom sessions. They were often called Notetakers and were brought into the room by our guests. These Notetakers were punctual—they were always on time—and remained quiet through the session.

It didn’t take long for us to realise they were AI bots that record and transcribe every word uttered during online meetings. Such bots have also started to attend offline meetings, through our phones and other recording devices.

As journalists, we found the trend fascinating. Not least because recording and transcribing had long been a part of our professional work. Earlier in our careers, we used to carry with us mini recorders, and additionally, a bunch of mini cassettes during our assignments. Often, we spent hours transcribing the interviews, typing them into MS Word documents (remembering to save every few minutes, because, if the computer crashed or the power went off, the entire work would be lost.)

The exercise was both useful (as it threw light on our mistakes, such as our tendency to interrupt too often, and revealed details that we failed to notice during the interview), and frustrating (to transcribe one hour of conversation typically took four to five hours).

So, we were also constantly looking for new technologies to help us. There weren't many. Even when we found one—Dragon Naturally Speaking, for example—it wasn't easy. We had to spend hours training it.

Now, AI has made it all easy, and we will share some of the tools we have found useful at the end of this piece. Before that, we want to share how we think about these tools, based on our experience as journalists. We are at the early stages of this phase of AI—there are new tools hitting the app stores every day, some of the existing ones are evolving fast, while some are dying. Thinking deeply about the underlying principles can help you choose better, based on your context.

Let's start with a broader philosophical idea, and then get down to specifics.

Our Extended Minds

Many of us fear technology, for good reasons. But that shouldn't stop us from trying them out. In fact, the tools are very much a part of ourselves, according to the so-called ‘extended mind’ theory. Its proponents, Andy Clark, a cognitive scientist and philosopher, and David Chalmers, a philosopher and consciousness researcher, argue that cognitive processes are not confined to the brain. They can extend into the external world through tools and technologies, like smartphones or notebooks, which act as integral parts of our mental systems. If a tool reliably aids in our memory or problem-solving skills, it should be considered part of the mind, they say. So, we should feel comfortable with tools because they enhance our intelligence, much like biological processes, and have historically been part of human evolution. They have enabled us to adapt and thrive in complex environments.

Check out this TED Talk by David Chalmers.

Once we get this basic idea, it's easy to see how these tools help us work better.

We Can Focus Better

When we take notes manually during meetings or interviews, it's a cognitive challenge. We attempt to listen, comprehend, contribute, and document all at the same time. All these can distract us from fully engaging with an ongoing discussion. When we “outsource” the recording to tech, it reduces the pressure on us to capture every detail in real time. We can use that mental space to listen well, ask good questions, and make thoughtful contributions.

We Can Make Meetings More Inclusive

Recording/transcribing might not sound like an exercise in inclusion. But, it could help, as we have discovered over the years. It’s especially true when a large team (in journalism, that usually means three or four people) is working on a big project. Not everyone can participate in all discussions for various reasons. Recordings and transcripts can keep everyone on the same page, to a large extent.

A Recorder Can Change Behaviour

As journalists, we have often seen people become more guarded when we turn the recorder on. It's almost always the case when we interview people for the first or second time, before the trust is established. However, once you have established the trust, recorders don't make a difference. It's important to keep this in mind. If people are uncomfortable with recording, work on building trust first.

Transcripts Matter

While audio or video recordings offer a complete capture of a conversation, reviewing them tends to take a lot of time. Transcripts are inherently more searchable, scannable, and manageable. This also increases the chance of your revisiting conversations.

Transcripts, essentially, democratise access to information within teams or organisations. Rather than depending on a single individual's notes, which may be incomplete or biased, or relying on collective memory, anyone in the team can independently verify decisions, recall specific details, or analyse content. 

Transcriptions can give fresh insights. Prof Saras Sarasvathy, a hugely respected expert in entrepreneurship, found the core ideas that formed the basis of her thesis, and the whole body of work that followed by poring over the transcripts of her interviews with entrepreneurs.

Context Matters

Often, recording conversations helps because participants are conscious of the permanence of their words, and offer deliberate and considered inputs. It can increase accountability.  

Sometimes, in more informal, creative, or brainstorming sessions, this same awareness could potentially be counterproductive if it leads to self-censorship or a reluctance to share nascent, less-polished ideas for fear of scrutiny.

In the pre-remote working era, the location took care of all these. When we discussed ideas in a coffee shop or office canteen, we were in a brainstorming mode. But when we were in the conference room discussing stories for the next issue, we tended to be more considered. When every meeting is virtual, we have to give more thought to the context and decide on recordings/transcriptions accordingly.

Archives Matter

When we were working on our book The Aadhaar Effect (Charles Assisi from Mumbai and NS Ramnath from Bangalore) we recorded all the interviews we did independently, and shared the transcripts with each other. But we also revisited the interviews we did much earlier with the Aadhaar team in the 2009-2010 period and found the material useful, because the consequences of some of the small decisions taken at that time became evident only much later. (For example, the early conversations the Aadhaar team had with professionals such as Nachiket Mor.) In business, people discard some ideas, not because ideas are bad, but because the time is not right. Transcriptions can help.

The AI Revolution

AI has transformed note-taking—from the quality of transcripts to the range of summaries. We would point to a piece Anmol Shrivastava wrote for Founding Fuel, and also the conversation he subsequently had with Tanuj Bhojwani. The lessons from that engagement hold good for this discussion too.

Check out the story: AI’s Best-Kept Secret: It’s Not the Tools, It’s You, and the conversation AMA: How to make AI work for you?.

Before we go on to the tools, it’s important to offer words of caution about recording and transcriptions. It's important to keep these in mind.

The Peltzman Effect

The awareness that a conversation is being recorded, and that a transcript will subsequently be available, can paradoxically lead to a reduction in active listening and real-time engagement. If we think we can simply "catch up later" by reviewing the recording or reading the transcript, our intrinsic motivation to pay close attention during the live session may diminish. This can result in less dynamic meetings, with fewer spontaneous contributions, less critical real-time discussion, and a generally more passive atmosphere.

It aligns closely with the Peltzman Effect. The theory suggests that individuals tend to adjust their behaviour in response to perceived changes in risk levels. When safety measures are introduced, making an environment or activity feel safer, people may behave less cautiously, thereby offsetting some of the intended safety benefits. In the context of meetings, the recording and the promise of a transcript act as such a “safety feature”. They reduce the perceived risk of missing crucial information or forgetting key decisions.

To counteract this, you may need to design it in a way that emphasises the value of real-time participation.

The Old Learning Tools

When you take notes, particularly by hand, it is a rich, multi-sensory learning experience. It involves the eyes, hands, and the physical sensation of hands moving on paper. All these apparently create stronger and more complex neural connections in the brain, enhancing our memory, deepening our comprehension. All journalists have at least once faced the tragedy of losing our recordings due to some technical glitch, so we have learnt by experience that, irrespective of the recording and the promise of 10-minute transcriptions, we still have to take notes.

While AI tools are getting better at mind maps, they are not as good as the ones you can scribble on paper as you listen to a discussion. So don't throw your pen and paper away.

A Question Of Ethics

Obviously, you have to take permission before recording conversations. Tools like Zoom announce it to participants by default. Yet, it's important to mention this explicitly.

Remember that in many jurisdictions it's a legal mandate that individuals must be informed and their explicit consent obtained before any conversation in which they are participating in is recorded. Recording an individual without their knowledge or consent is a breach of personal privacy and interpersonal trust. Such actions can undermine open communication, damage relationships, and potentially lead to legal repercussions, depending on the applicable laws.

In journalism, when recording a conversation, we used to do the elaborate ritual of switching the recorder on and placing it in front of the person we were interviewing, making it very clear that we were recording. Do the equivalent of that whenever you record.

Large Language Models (LLMs) pose a new ethical challenge because when we share the transcript with an LLM for summary or search, it often becomes a part of the LLM’s training material. Some might be okay with your recording and transcribing the meetings, but might be uncomfortable with your sharing it with ChatGPT or Claude or Gemini.

Data Ownership

This also brings “data ownership” into focus. If your conversation, even if de-identified, contributes to the training and improvement of a commercial LLM that subsequently generates significant economic value for the AI company, what about your residual rights or claims? If your data helped create or refine a model, do you have a stake in the value generated, or do you have any control over how your informational contribution is used in downstream applications of the AI? These are largely unresolved legal and ethical questions that lie at the frontier of AI governance and data rights.

We need more discussions on that. But it's good to remember that obtaining basic consent to record a conversation is only one level of permission. Eventually, we might have to take multiple levels of consent, including for transcripts, summaries, analytical insights, or metadata and how they will be handled.

These are important questions, and we need more conversations. Sometimes these discussions take us into the realm of philosophy. Which only highlights the fact that it's a hard problem.

Personal Assistants

Philosophers and thinkers like Socrates or Ramakrishna Paramahamsa didn't believe in writing things down. They were rooted in oral discourse, through asking and answering questions in live conversation. Yet, we have access to their wisdom because of people like Plato and Mahendranath Gupta, who transcribed what they said. We might not be a Socrates or Ramakrishna, but we could get AI to do what Plato and M (as Mahendranath Gupta’s name appears in The Gospel of Sri Ramakrishna) did. The work might not change the world, but it could be useful for us.

Tools We Have Tried

Otter.ai

Who It's For

• Professionals who attend many meetings
• Teams that need accurate meeting records
• People who collaborate remotely

Potential Use Cases

• Recording and transcribing team meetings
• Creating searchable meeting archives
• Assigning and tracking action items from discussions

What to Keep in Mind

• Your conversations help train their AI systems
• The free plan has limited minutes per month
• Transcripts may contain personal information

Difficulty Level

• Beginner—easy to set up and use

Other Points

• Integrates with common video meeting platforms
• Offers speaker identification
• Can handle industry-specific terms with custom vocabularies

MacWhisper

Who It's For

• Privacy-conscious Mac users
• People who need offline transcription
• Users who work with varied audio formats

Potential Use Cases

• Transcribing interviews or research recordings
• Creating subtitles for videos
• Converting voice notes to text

What to Keep in Mind

• More accurate models require paid upgrades
• Only works on Mac devices
• Processing speed depends on your computer

Difficulty Level

• Beginner to Intermediate—simple interface but some options require learning

Other Points

• All processing happens on your device
• No internet connection needed for transcription
• Supports multiple languages

SuperWhisper

Who It's For

• Mac users who dictate frequently
• People who need customised transcription settings
• Users concerned about privacy

Potential Use Cases

• Dictating documents or emails
• Converting existing audio/video files to text
• Translating spoken content between languages

What to Keep in Mind

• Uses subscription payment model
• Some features may require your own API keys
• Requires macOS or iOS

Difficulty Level

• Intermediate—offers advanced features that take time to learn

Other Points

• Can adapt to different applications you’re using
• No server storage of your content
• Option to connect to additional AI services

Talktastic

Who It's For

• Mac users who dictate across different applications
• People who want context-aware dictation
• Users who need both raw and improved transcripts

Potential Use Cases

• Dictating emails while seeing previous messages
• Creating content that matches the style of surrounding text
• Taking quick voice notes with automatic cleanup

What to Keep in Mind

• Currently in beta testing
• Takes screenshots to understand context
• Future pricing not yet announced

Difficulty Level

• Beginner—designed to be straightforward

Other Points

• Shows both original and AI-improved versions
• Uses screen content to improve accuracy
• Still developing privacy features like encryption

Google Pinpoint

Who It's For

• Journalists researching stories
• Academic researchers
• People working with large document collections

Potential Use Cases

• Analyzing interview transcripts
• Finding patterns across many documents
• Converting handwritten notes to searchable text

What to Keep in Mind

• Only available to verified journalists and researchers
• Uploads are subject to Google’s terms
• Grants Google permission to use your content

Difficulty Level

• Intermediate—powerful features require some learning

Other Points

• Can handle thousands of documents
• Identifies key people and organisations automatically
• Allows collaboration with team members

All of these transcription tools offer free versions or trial periods, which we strongly recommend trying before committing to any purchases. This hands-on experience helps you determine which tool best fits your specific needs and workflow.

For example, one of us (Ramnath) tested several options for a few weeks, before deciding to invest in lifetime licenses for both MacWhisper ($21.4 for a single device) and SuperWhisper ($149 for multiple devices including mobile phone, during a sale) as they matched requirements.

Your ideal choice will depend entirely on your usage patterns, privacy concerns, and the specific features you value most.

We would love to hear which transcription tools you’ve found most useful in your work. Please share your experiences and recommendations in the comments!